(M1) (Raymond Barnett) Finite Mathematics For Business

MyLab Math for Finite Mathematics
for Business, Economics, Life

Sciences, and Social Sciences, 14e
(access code required)
Used by over 3 million students a year, MyLab™ Math is the world’s leading online
program for teaching and learning mathematics. MyLab Math delivers assessment,
tutorials, and multimedia resources that provide engaging and personalized
experiences for each student, so learning can happen in any environment.
Interactive Figures
A full suite of Interactive Figures has been added to
support teaching and learning. The figures illustrate
key concepts and allow manipulation. They have been
designed to be used in lecture as well as by students
independently.
Questions that Deepen Understanding

MyLab Math includes a variety of question types
designed to help students succeed in the course. In
Setup & Solve questions, students show how they set
up a problem as well as the solution, better mirroring
what is required on tests. Additional Conceptual
Questions provide support for assessing concepts
and vocabulary. Many of these questions are
application oriented.
pearson.com/mylab/math
fourteenth edition
FINITE
MATHEMATICS
for Business, Economics,
Life Sciences, and
Social Sciences
GLOBAL EDITION
RAYMOND A. BARNETT Merritt College
MICHAEL R. ZIEGLER Marquette University
KARL E. BYLEEN Marquette University
CHRISTOPHER J. STOCKER Marquette University

Director, Portfolio Management: Deirdre Lynch Product Marketing Manager: Emily Ockay
Executive Editor: Jeff Weidenaar Field Marketing Manager: Evan St. Cyr
Editorial Assistant: Jennifer Snyder Marketing Assistants: Erin Rush, Shannon McCormack
Acquisitions Editor, Global Edition: Sourabh Maheshwari Senior Author Support/Technology Specialist: Joe Vetere
Assistant Project Editor, Global Edition: Sulagna Dasgupta Manager, Rights and Permissions: Gina Cheselka
Content Producers: Sherry Berg, Ron Hampton Manufacturing Buyer: Carol Melville, LSC Communications
Managing Producer: Karen Wernholm Senior Manufacturing & Composition Controller, Global Edition:
Senior Producer: Stephanie Green Angela Hawksbee
Media Production Manager, Global Edition: Vikram Kumar Art Director: Barbara Atkinson
Manager, Courseware QA: Mary Durnwald Production Coordination, Composition, and Illustrations: Integra
Manager, Content Development: Kristina Evans
Cover Image: Mark Nazh/Shutterstock

Photo Credits: Page 19: acro_phuket/Shutterstock; Page 104: IOvE IOvE/Shuttertock; Page 152: Fedetov Anatoly/Shutterstock; Page 235:
Ramann/Shutterstock; Page 265: Fotosenmeer/Shutterstock; Page 329: PR Image Factory/Shutterstock; Page 372: noomcpk/
Shutterstock; Page 436: Taaee/123RF; Page 478: Franck Boston/Shutterstock; G1: Zeljko Radojko/Shutterstock; Page 530: Durk
Talsma/Shuttertock
Text Credits: Page 477: American Lung Association; Page 479: Federal Aviation Administration; Page 479: U.S. Treasury; Page 480: Institute
of Education Sciences; Page 480: UN Population Division; Page 487: American Petroleum Institute; Page 487: Fortune; Page
487: U. S . Postal Service; Page 487: U.S. Bureau of Economic Analysis; Page 487: U.S. Geological Survey; Page 487: U.S.
Treasury; Page 498: American Petroleum Institute; Page 498: The Institute for College Access & Success; Page 498: World
Tourism Organization; Page 499: U.S. Census Bureau; Page 527: U.S. Elections Project; Page 538: U.S. Bureau of Labor Statistics;
Page 553: biggreenegg.com; Page 557: www.tradeshop.com; Page 558: www.tradeshop.com; Page 559: National Oceanic and
Atmospheric Administration; Page 560: Jack Haggerty Forest, Lakehead University, Canada; Page 562: Energy Information
Administration; Page 563: Bureau of Transportation Statistics; Page 563: Bureau of Transportation Statistics; Page 563: Centers for
Disease Control; Page 563: Walmart Stores, Inc.; Page 564: Kagen Research; Page 564: Lakehead University; Page 564: T. Labuza,
University of Minnesota; Page 565: Infoplease.com; Page 565: National Center for Education Statistics; Page 565: National
Oceanic and Atmospheric Administration; Page 566: www.usda.gov/nass/pubslhistdata.htm; Page 566: Ashland Inc.
Pearson Education Limited

KAO Two
KAO Park
Harlow
CM17 9NA
United Kingdom
and Associated Companies throughout the world
Visit us on the World Wide Web at: www.pearsonglobaleditions.com
© Pearson Education Limited 2019
The rights of Raymond A. Barnett, Michael R. Ziegler, Karl E. Byleen, and Christopher J. Stocker to be identified as the authors of this work
have been asserted by them in accordance with the Copyright, Designs and Patents Act 1988.
Authorized adaptation from the United States edition, entitled Finite Mathematics for Business, Economics, Life Sciences, and Social Sciences,
14th Edition, ISBN 978-0-134-67598-5 by Raymond A. Barnett, Michael R. Ziegler, Karl E. Byleen, Christopher J. Stocker, published by
Pearson Education © 2019.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, elec-
tronic, mechanical, photocopying, recording or otherwise, without either the prior written permission of the publisher or a license permitting re-
stricted copying in the United Kingdom issued by the Copyright Licensing Agency Ltd, Saffron House, 6–10 Kirby Street, London EC1N 8TS.
All trademarks used herein are the property of their respective owners. The use of any trademark in this text does not vest in the author or pub-
lisher any trademark ownership rights in such trademarks, nor does the use of such trademarks imply any affiliation with or endorsement of this
book by such owners. For information regarding permissions, request forms, and the appropriate contacts within the Pearson Education Global
Rights and Permissions department, please visit www.pearsoned.com/permissions.
This eBook is a standalone product and may or may not include all assets that were part of the print version. It also does not provide access to
other Pearson digital products like MyLab and Mastering. The publisher reserves the right to remove any material in this eBook at any time.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
ISBN 10: 1-292-26420-9
ISBN 13: 978-1-292-26420-2
eBook ISBN 13: 978-1-292-26429-5
Typeset by Integra Software Services Private Limited
CONTENTS
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Diagnostic Prerequisite Test . . . . . . . . . . . . . . . . . . . . 17
Chapter 1 Functions and Graphs . . . . . . . . . . . . . . . . . . . 19

1.1 Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . .20
1.2 Elementary Functions: Graphs and Transformations . . . . . . . .34
1.3 Quadratic Functions . . . . . . . . . . . . . . . . . . . . . .46
1.4 Polynomial and Rational Functions . . . . . . . . . . . . . . . .62
1.5 Exponential Functions . . . . . . . . . . . . . . . . . . . . . .74
1.6 Logarithmic Functions . . . . . . . . . . . . . . . . . . . . . .84
Chapter 1 Summary and Review . . . . . . . . . . . . . . . . . . .96
Review Exercises . . . . . . . . . . . . . . . . . . . . . . . . . .98
Chapter 2 Mathematics of Finance . . . . . . . . . . . . . . . . . . 104

2.1 Simple Interest . . . . . . . . . . . . . . . . . . . . . . . . 105
2.2 Compound and Continuous Compound Interest . . . . . . . . . 112
2.3 Future Value of an Annuity; Sinking Funds. . . . . . . . . . . . 126
2.4 Present Value of an Annuity; Amortization . . . . . . . . . . . 134
Chapter 2 Summary and Review . . . . . . . . . . . . . . . . . . 146
Review Exercises . . . . . . . . . . . . . . . . . . . . . . . . . 148
Chapter 3 Systems of Linear Equations; Matrices . . . . . . . . . . . . 152

3.1 Review: Systems of Linear Equations in Two Variables . . . . . . 153
3.2 Systems of Linear Equations and Augmented Matrices . . . . . . 166
3.3 Gauss–Jordan Elimination . . . . . . . . . . . . . . . . . . . 175
3.4 Matrices: Basic Operations . . . . . . . . . . . . . . . . . . 189
3.5 Inverse of a Square Matrix . . . . . . . . . . . . . . . . . . 202
3.6 Matrix Equations and Systems of Linear Equations. . . . . . . . 214
3.7 Leontief Input–Output Analysis . . . . . . . . . . . . . . . . . 222
Chapter 4 Linear Inequalities and Linear Programming . . . . . . . . . 235

4.1 Linear Inequalities in Two Variables . . . . . . . . . . . . . . 236
4.2 Systems of Linear Inequalities in Two Variables . . . . . . . . . 244
4.3 Linear Programming in Two Dimensions: A Geometric Approach . 250
Chapter 5 Linear Programming: The Simplex Method . . . . . . . . . . 265

5.1 The Table Method: An Introduction to the Simplex Method . . . . 266
5.2 The Simplex Method: Maximization with Problem
Constraints of the Form … . . . . . . . . . . . . . . . . . . 277
3
4 CONTENTS
5.3 The Dual Problem: Minimization with Problem

Constraints of the Form Ú . . . . . . . . . . . . . . . . . . . 294
5.4 Maximization and Minimization with Mixed Problem
Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 308
Chapter 6 Logic, Sets, and Counting . . . . . . . . . . . . . . . . . 329

6.1 Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
6.2 Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
6.3 Basic Counting Principles . . . . . . . . . . . . . . . . . . . 346
6.4 Permutations and Combinations . . . . . . . . . . . . . . . . 355
Chapter 7 Probability . . . . . . . . . . . . . . . . . . . . . . . . . 372

7.1 Sample Spaces, Events, and Probability . . . . . . . . . . . . 373
7.2 Union, Intersection, and Complement of Events; Odds . . . . . . 386
7.3 Conditional Probability, Intersection, and Independence . . . . . 399
7.4 Bayes’ Formula . . . . . . . . . . . . . . . . . . . . . . . . 413
7.5 Random Variable, Probability Distribution, and Expected Value . . 421
Chapter 8 Markov Chains . . . . . . . . . . . . . . . . . . . . . . 436

8.1 Properties of Markov Chains. . . . . . . . . . . . . . . . . . 437
8.2 Regular Markov Chains . . . . . . . . . . . . . . . . . . . . 448
8.3 Absorbing Markov Chains. . . . . . . . . . . . . . . . . . . 459
Chapter 9 Data Description and Probability Distributions . . . . . . . . 478

9.1 Graphing Data . . . . . . . . . . . . . . . . . . . . . . . . 479
9.2 Measures of Central Tendency . . . . . . . . . . . . . . . . . 490
9.3 Measures of Dispersion . . . . . . . . . . . . . . . . . . . . 500
9.4 Bernoulli Trials and Binomial Distributions. . . . . . . . . . . . 506
9.5 Normal Distributions . . . . . . . . . . . . . . . . . . . . . 516
Chapter 10 Games and Decisions (online) . . . . . . . . . . . . . . . G1

10.1 Strictly Determined Games . . . . . . . . . . . . . . . . . . G2
10.2 Mixed-Strategy Games. . . . . . . . . . . . . . . . . . . . G8
10.3 Linear Programming and 2 * 2 Games: A Geometric
Approach. . . . . . . . . . . . . . . . . . . . . . . . . . G19
10.4 Linear Programming and m * n Games: Simplex Method
and the Dual Problem . . . . . . . . . . . . . . . . . . . . G25
CONTENTS 5
Chapter 10 Summary and Review . . . . . . . . . . . . . . . . . G31

Review Exercises . . . . . . . . . . . . . . . . . . . . . . . . . G32
Appendix A Linear Equations and Graphs . . . . . . . . . . . . . . . . 530

A.1 Linear Equations and Inequalities . . . . . . . . . . . . . . . 531
A.2 Graphs and Lines . . . . . . . . . . . . . . . . . . . . . . . 541
A.3 Linear Regression . . . . . . . . . . . . . . . . . . . . . . . 555
Appendix B Basic Algebra Review . . . . . . . . . . . . . . . . . . . 567

B.1 Real Numbers . . . . . . . . . . . . . . . . . . . . . . . . 567
B.2 Operations on Polynomials . . . . . . . . . . . . . . . . . . 573
B.3 Factoring Polynomials . . . . . . . . . . . . . . . . . . . . . 579
B.4 Operations on Rational Expressions . . . . . . . . . . . . . . 585
B.5 Integer Exponents and Scientific Notation . . . . . . . . . . . 591
B.6 Rational Exponents and Radicals. . . . . . . . . . . . . . . . 595
B.7 Quadratic Equations . . . . . . . . . . . . . . . . . . . . . 601
Appendix C Special Topics (online) . . . . . . . . . . . . . . . . . . . A1

C.1 Sequences, Series, and Summation Notation . . . . . . . . . . A1
C.2 Arithmetic and Geometric Sequences. . . . . . . . . . . . . . A7
C.3 Binomial Theorem. . . . . . . . . . . . . . . . . . . . . . . A13
Appendix D Area under the Standard Normal Curve . . . . . . . . . . 610

Answers . . . . . . . . . . . . . . . . . . . . . . . . . . A-1
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . .I-1
Index of Applications . . . . . . . . . . . . . . . . . . . . I-10
PREFACE
The fourteenth edition of Finite Mathematics for Business, Economics, Life Sciences,
and Social Sciences is designed for a one-term course in finite mathematics for stu-
dents who have had one to two years of high school algebra or the equivalent. The
book’s overall approach, refined by the authors’ experience with large sections of col-
lege freshmen, addresses the challenges of teaching and learning when prerequisite
knowledge varies greatly from student to student.
Note that Chapters 1–8 of this text also appear in College Mathematics for
Business, Economics, Life Sciences, and Social Sciences, by the same author team.
The College Mathematics text also contains coverage of applied calculus topics.
The authors had three main goals in writing this text:
1. To write a text that students can easily comprehend
2. To make connections between what students are learning and how they may
apply that knowledge
3. To give flexibility to instructors to tailor a course to the needs of their students.
Many elements play a role in determining a book’s effectiveness for students. Not
only is it critical that the text be accurate and readable, but also, in order for a book
to be effective, aspects such as the page design, the interactive nature of the presenta-
tion, and the ability to support and challenge all students have an incredible impact
on how easily students comprehend the material. Here are some of the ways this text
addresses the needs of students at all levels:
■■ Page layout is clean and free of potentially distracting elements.
■■ Matched Problems that accompany each of the completely worked examples
help students gain solid knowledge of the basic topics and assess their own level
of understanding before moving on.
■■ Review material (Appendix A, B, and Chapter 1) can be used judiciously to help
remedy gaps in prerequisite knowledge.
■■ A Diagnostic Prerequisite Test prior to Chapter 1 helps students assess their
skills, while the Basic Algebra Review in Appendix B provides students with the
content they need to remediate those skills.
■■ Explore and Discuss problems lead the discussion into new concepts or build
upon a current topic. They help students of all levels gain better insight into the
mathematical concepts through thought-provoking questions that are effective in
both small and large classroom settings.
■■ Instructors are able to easily craft homework assignments that best meet the
needs of their students by taking advantage of the variety of types and dif-
ficulty levels of the exercises. Exercise sets at the end of each section consist
of a Skills Warm-up (four to eight problems that review prerequisite knowl-
edge specific to that section) followed by problems divided into categories
A, B, and C by level of difficulty, with level-C exercises being the most
challenging.
■■ The MyLab Math course for this text is designed to help students help them-
selves and provide instructors with actionable information about their progress.
The immediate feedback students receive when doing homework and practice
in MyLab Math is invaluable, and the easily accessible eText enhances student
learning in a way that the printed page sometimes cannot.
■■ Most important, all students get substantial experience in modeling and solving
real-world problems through application examples and exercises chosen from
6
PREFACE 7
business and economics, life sciences, and social sciences. Great care has been
taken to write a book that is mathematically correct, with its emphasis on com-
putational skills, ideas, and problem solving rather than mathematical theory.
■■ Finally, the choice and independence of topics make the text readily adaptable
to a variety of courses.
New to This Edition

Fundamental to a book’s effectiveness is classroom use and feedback. Now in its
fourteenth edition, this text has had the benefit of a substantial amount of both.
Improvements in this edition evolved out of the generous response from a large num-
ber of users of the last and previous editions as well as survey results from instruc-
tors. Additionally, we made the following improvements in this edition:
■■ Redesigned the text in full color to help students better use it and to help moti-
vate students as they put in the hard work to learn the mathematics (because let’s
face it—a more modern looking book has more appeal).
■■ Updated graphing calculator screens to TI-84 Plus CE (color) format.
■■ Added Reminder features in the side margin to either remind students of a con-
cept that is needed at that point in the book or direct the student back to the sec-
tion in which it was covered earlier.
■■ Updated data in examples and exercises. Many modern and student-centered
applications have been added to help students see the relevance of the content.
■■ Analyzed aggregated student performance data and assignment frequency data
from MyLab Math for the previous edition of this text. The results of this analy-
sis helped improve the quality and quantity of exercises that matter the most to
instructors and students.
■■ Rewrote and simplified the treatment of cost, revenue, and profit in Section 1.1.
■■ Added 611 new exercises throughout the text.
■■ Moved the seldom-used chapter “Games and Decisions” online to
www.pearsonglobaleditions.com. Note that all of the resources that for-
merly accompanied this chapter are still available. They are housed within
MyLab Math.
New to MyLab Math

Many improvements have been made to the overall functionality of MyLab Math
since the previous edition. However, beyond that, we have also increased and
improved the content specific to this text.
■■ Instructors now have more exercises than ever to choose from in assigning home-
work. Most new questions are application-oriented. There are approximately 3,340
assignable exercises in MyLab Math for this text. New exercise types include:
●■ Additional Conceptual Questions provide support for assessing concepts
and vocabulary. Many of these questions are application-oriented.
●■ Setup & Solve exercises require students to show how they set up a problem
as well as the solution, better mirroring what is required of students on tests.
■■ The Guide to Video-Based Assignments shows which MyLab Math exercises
can be assigned for each video. (All videos are also assignable.) This resource is
handy for online or flipped classes.
■■ The Note-Taking Guide provides support for students as they take notes in class.
The Guide includes definitions, theorems, and statements of examples but has blank
space for students to write solutions to examples and sample problems. The Note-
Taking Guide corresponds to the Lecture PowerPoints that accompany the text. The
Guide can be downloaded in PDF or Word format from within MyLab Math.
8 PREFACE
■■ A full suite of Interactive Figures has been added to support teaching and
learning. The figures illustrate key concepts and allow manipulation. They have
been designed to be used in lecture as well as by students independently.
■■ Study Skills Modules help students with the life skills that can make the differ-
ence between passing and failing.
■■ MathTalk and StatTalk videos highlight applications of the content of the
course to business. The videos are supported by assignable exercises.
■■ The Graphing Calculator Manual and Excel Spreadsheet Manual, both spe-
cific to this course, have been updated to support the TI-84 Plus CE (color edi-
tion) and Excel 2016, respectively. Both manuals also contain additional topics
to support the course. These manuals are within the Tools for Success tab.
■■ MyLab Math now contains a downloadable Instructor’s Answers document—
with all answers in one place. (This augments the downloadable Instructor’s
Solutions Manual, which contains even-numbered solutions.)
Trusted Features
■■ Emphasis and Style—As was stated earlier, this text is written for student
comprehension. To that end, the focus has been on making the book both math-
ematically correct and accessible to students. Most derivations and proofs are
omitted, except where their inclusion adds significant insight into a particular
concept as the emphasis is on computational skills, ideas, and problem solving
rather than mathematical theory. General concepts and results are typically pre-
sented only after particular cases have been discussed.
■■ Design—One of the hallmark features of this text is the clean, straightforward
design of its pages. Navigation is made simple with an obvious hierarchy of
key topics and a judicious use of call-outs and pedagogical features. A func-
tional use of color improves the clarity of many illustrations, graphs, and ex-
planations, and guides students through critical steps (see pages 37 and 38).
■■ Examples—More than 300 completely worked examples are used to introduce
concepts and to demonstrate problem-solving techniques. Many examples have
multiple parts, significantly increasing the total number of worked examples.
The examples are annotated using blue text to the right of each step, and the
problem-solving steps are clearly identified. To give students extra help in work-
ing through examples, dashed boxes are used to enclose steps that are usually per-
formed mentally and rarely mentioned in other books (see Example 7 on page 29).
Though some students may not need these additional steps, many will appreciate
the fact that the authors do not assume too much in the way of prior knowledge.
■■ Matched Problems—Each example is followed by a similar Matched Problem
for the student to work while reading the material. This actively involves the
student in the learning process. The answers to these matched problems are
included at the end of each section for easy reference.
■■ Explore and Discuss—Most every section contains Explore and Discuss prob-
lems at appropriate places to encourage students to think about a relationship
or process before a result is stated or to investigate additional consequences of
a development in the text (see pages 36 and 39). This serves to foster critical
thinking and communication skills. The Explore and Discuss material can be
used for in-class discussions or out-of-class group activities and is effective in
both small and large class settings.
■■ Exercise Sets—The book contains over 4,200 carefully selected and graded
exercises. Many problems have multiple parts, significantly increasing the
total number of exercises. Writing exercises, indicated by the icon , provide
PREFACE 9
students with an opportunity to express their understanding of the topic in

writing. Answers to all odd-numbered problems are in the back of the book.
Exercises are paired so that consecutive odd- and even-numbered exercises are
of the same type and difficulty level. Exercise sets are structured to facilitate
crafting just the right assignment for students:
●■ Skills Warm-up exercises, indicated by W , review key prerequisite knowledge.
●■ Graded exercises: Levels A (routine, easy mechanics), B (more difficult
mechanics), and C (difficult mechanics and some theory) make it easy for
instructors to create assignments that are appropriate for their classes.
●■ Applications conclude almost every exercise set. These exercises are labeled
with the type of application to make it easy for instructors to select the right
exercises for their audience.
■■ Applications—A major objective of this book is to give the student substantial
experience in modeling and solving real-world problems. Enough applications
are included to convince even the most skeptical student that mathematics is
really useful (see the Index of Applications at the back of the book). Almost
every exercise set contains application problems, including applications from
business and economics, life sciences, and social sciences. An instructor with
students from all three disciplines can let them choose applications from their
own field of interest; if most students are from one of the three areas, then
special emphasis can be placed there. Most of the applications are simplified
versions of actual real-world problems inspired by professional journals and
books. No specialized experience is required to solve any of the application
problems.
■■ Graphing Calculator and Spreadsheets—Although access to a graphing
calculator or spreadsheets is not assumed, it is likely that many students will
want to make use of this technology. To assist these students, optional graph-
ing calculator and spreadsheet activities are included in appropriate places.
These include brief discussions in the text, examples or portions of examples
solved on a graphing calculator or spreadsheet, and exercises for the students
to solve. For example, regression techniques on a graphing calculator are
used at appropriate points to illustrate mathematical modeling with real data.
All the optional graphing calculator material is clearly identified with the
icon and can be omitted without loss of continuity, if desired. Graphing
calculator screens displayed in the text are actual output from the TI-84 Plus
CE (color version) graphing calculator.
Additional Pedagogical Features

The following features, while helpful to any student, are particularly helpful to stu-
dents enrolled in a large classroom setting where access to the instructor is more
challenging or just less frequent. These features provide much-needed guidance for
students as they tackle difficult concepts.
■■ Call-out boxes highlight important definitions, results, and step-by-step pro-
cesses (see pages 40, 47, and 105).
■■ Caution statements appear throughout the text where student errors often occur
(see pages 28 and 93).
■■ Conceptual Insights, appearing in nearly every section, often make explicit
connections to previous knowledge but sometimes encourage students to think
beyond the particular skill they are working on and attain a more enlightened
view of the concepts at hand (see pages 36 and 47).
10 PREFACE
■■ Diagnostic Prerequisite Test, located on page 17, provides students with a

tool to assess their prerequisite skills prior to taking the course. The Basic
Algebra Review, in Appendix B, provides students with seven sections of con-
tent to help them remediate in specific areas of need. Answers to the Diagnostic
Prerequisite Test are at the back of the book and reference specific sections in the
Appendix A, B, and Chapter 1 for students to use for remediation.
■■ Chapter Reviews—Often it is during the preparation for a chapter exam that
concepts gel for students, making the chapter review material particularly im-
portant. The chapter review sections in this text include a comprehensive sum-
mary of important terms, symbols, and concepts, keyed to completely worked
examples, followed by a comprehensive set of Review Exercises. Answers to
Review Exercises are included at the back of the book; each answer contains a
reference to the section in which that type of problem is discussed so students
can remediate any deficiencies in their skills on their own.
Content
The text begins with the development of a library of elementary functions in
Chapter 1, including their properties and applications. Many students willbe familiar
with most, if not all, of the material in this introductory chapter. Depending on stu-
dents’ preparation and the course syllabus, an instructor has several options for using
the first chapter and appendices including the following:
(i) Skip Chapter 1 and refer to it only as necessary later in the course;
(ii) Cover Appendix A quickly in the first week of the course, emphasizing price–de-
mand equations, price–supply equations, and linear regression, but skip Chapter 1;
(iii) Cover Appendix A and Chapter 1 systematically before moving on to other chapters.
The material in Finite Mathematics can be thought of as four units:
1. Mathematics of finance (Chapter 2)
2. Linear algebra, including matrices, linear systems, and linear programming
(Chapters 3, 4, and 5)
3. Probability and statistics (Chapters 6 and 7)
4. Applications of linear algebra and probability to Markov chains and game theory
(Chapters 8 and 10)
The first three units are independent of each other, while the fourth unit is dependent
on some of the earlier chapters. (Markov chains requires Chapters 3 and 7; game
theory requires Chapters 3–5 and 7).
Chapter 2 presents a thorough treatment of simple and compound interest and pres-
ent and future value of ordinary annuities. Appendix C.1 addresses arithmetic and
geometric sequences and can be covered in conjunction with this chapter, if desired.
Chapter 3 covers linear systems and matrices with an emphasis on using row op-
erations and Gauss–Jordan elimination to solve systems and to find matrix inverses.
This chapter also contains numerous applications of mathematical modeling using
systems and matrices. To assist students in formulating solutions, all answers at the
back of the book for application exercises in Sections 3.3, 3.5, and the chapter Review
Exercises contain both the mathematical model and its solution. The row operations
discussed in Sections 3.2 and 3.3 are required for the simplex method in Chapter 5.
Matrix multiplication, matrix inverses, and systems of equations are required for
Markov chains in Chapter 8.
Chapters 4 and 5 provide a broad and flexible coverage of linear programming.
Chapter 5 covers two-variable graphing techniques. Instructors who wish to emphasize
linear programming techniques can cover the basic simplex method in Sections 5.1
and 5.2 and then discuss either or both of the following: the dual method (Section 5.3)
PREFACE 11
and the big M method (Section 5.4). Those who want to emphasize modeling can dis-
cuss the formation of the mathematical model for any of the application examples in
Sections 5.2–5.4, and either omit the solution or use software to find the solution. To
facilitate this approach, all answers at the back of the book for application exercises
in Sections 5.2–5.4 and the chapter Review Exercises contain both the mathematical
model and its solution. The simplex and dual solution methods are required for por-
tions of Chapter 10.
Chapter 6 provides a foundation for probability with a treatment of logic, sets, and
counting techniques.
Chapter 7 covers basic probability, including Bayes’ formula and random variables.
Chapter 9 deals with basic descriptive statistics and more advanced probability dis-
tributions, including the important normal distribution. Appendix C.3 contains a short
discussion of the binomial theorem that can be used in conjunction with the develop-
ment of the binomial distribution in Section 10.4.
Chapters 8 and 10 tie together concepts developed in earlier chapters and apply them
to interesting topics. A study of Markov chains (Chapter 8) or game theory (Chapter
10) provides an excellent unifying conclusion to a finite mathematics course.
Appendix B contains a concise review of basic algebra that may be covered as part
of the course or referenced as needed. As mentioned previously, Appendix C (online
at www.pearsonglobaleditions.com) contains additional topics that can be covered in
conjunction with certain sections in the text, if desired.
Acknowledgments
In addition to the authors, many others are involved in the successful publication of a
book. We wish to thank the following reviewers:
Ben Anilus, Roger Williams University
Paula F. Bailey, Roger Williams University
Mary Ann Barber, University of North Texas
Sandra Belcher, Midwestern State University
Chris Bendixen, Lake Michigan College
Brian C. Bright, Ivy Tech Community College
Ronnie L. Brown, American University
Dee Cochran, Indiana University, Southeast
Lauren Fern, University of Montana
Sonya Geisel, Arizona State University
Jeffrey Hakim, American University
Patricia Ann Hussey, Triton College
Chuck Jackson, Lone Star College, Kingwood
Sheila C. Luke, Colorado Technical University
Mari M. Menard, Lone Star College, Kingwood
Faramarz Mokhtari, Montgomery College, Rockville
Jimmy Mopecha, Delaware Valley University
Luis Molina, Lone Star College, CyFair
Michael Puente, Richland College
Rick Pugsley, Ivy Tech Community College
Douglas Ray, Texas State University
Frances Smith, Richland College
John C. Treacy, Ivy Tech Community College
The following faculty members provided direction on the development of the MyLab
Math course for this edition:
Mary Ann Barber, University of North Texas
Ronnie L. Brown, University of Baltimore
12 PREFACE
David Busekist, Southeastern Louisiana University

Kristen A. Ceballos, Southern Illinois University, Carbondale
Brandi L. Cline, Lone Star College, Tomball
James Colton, Shaker High School
Natalia Gavryshova, College of San Mateo
Liz Hawk, Eastbrook High School
Natalie Herndon, Southwest Tennessee State Univerity
Alan Kunkle, Ivy Tech Community College, Kokomo
Steve Lam, Guam Community College
Mari M. Menard, Lone Star College, Kingwood
Faramarz Mokhtari, Montgomery College, Rockville
Jimmy Mopecha, Delaware Valley University
Mike Panahi, El Centro College
Terri Seiver, San Jacinto College
Jack Wagner, Fordham University
Dennis Wolf, Indiana University, South Bend
Dinesh S. Yadav, Richland College
We also express our thanks to John Samons and Damon Demas for providing a care-
ful and thorough accuracy check of the text, problems, and answers. Our thanks to
Garret Etgen, John Samons, Salvatore Sciandra, Victoria Baker, Ben Rushing, and
Stela Pudar-Hozo for developing the supplemental materials so important to the suc-
cess of a text. And finally, thanks to all the people at Pearson and Integra who con-
tributed their efforts to the production of this book.
Acknowledgments for the Global Edition

Pearson would like to thank the following people for their contributions to the Global
Edition of this book.
Contributors
Veena Dhingra
Katarzyna Zuleta Estrugo, École Polytechnique Fédérale de Lausanne
Reviewers
Sunil Jacob John, National Institute of Technology Calicut
Mohammad Kacim, Holy Spirit University of Kaslik
Seifedine Kadry, Beirut Arab University
M. Sankar, Presidency University
We also express our gratitude to Vinod Ramachandran and Tibor Vörös for providing
their feedback on how to enhance this book by participating in the survey.
PREFACE 13
MyLab Math Online Course

for Finite Mathematics for Business,
Economics, Life Sciences, and
Social Sciences, 14e (access code required)
MyLabTM Math is available to accompany Pearson’s market-leading text offerings. To give students a
consistent tone, voice, and teaching method, each text’s flavor and approach are tightly integrated
throughout the accompanying MyLab Math course, making learning the material as seamless as possible.
Preparedness
One of the biggest challenges in applied math courses is making sure students are adequately
prepared with the prerequisite skills needed to successfully complete their course work. MyLab Math
supports students with just-in-time remediation and key-concept review.
NEW! Study Skills Modules

Study skills modules help students with the life skills that can make the difference between passing and
failing.
Developing Deeper Understanding

MyLab Math provides content and
tools that help students build a deep-
er understanding of course content
than would otherwise be possible.
Exercises with
Immediate Feedback
Homework and practice exercises for
this text regenerate algorithmically to
give students unlimited opportunity
for practice and mastery. MyLab
Math provides helpful feedback when
students enter incorrect answers and
includes the optional learning aids
Help Me Solve This, View an Exam-
ple, videos, and/or the eText.
14 PREFACE
NEW! Additional Conceptual Questions

Additional Conceptual Questions provide support for assessing concepts and vocabulary. Many of these
questions are application-oriented. They are clearly labeled “Conceptual” in the Assignment Manager.
NEW! Setup & Solve Exercises

These exercises require students to show how they
set up a problem as well as the solution, better mir-
roring what is required on tests.
NEW! Interactive Figures

A full suite of Interactive Figures has been added to support
teaching and learning. The figures illustrate key concepts and
allow manipulation. They are designed to be used in lecture
as well as by students independently.
Instructional Videos
Every example in the text has an instructional video tied to it that can be used as a learning aid or for
self-study. MathTalk videos were added to highlight business applications to the course content, and a
Guide to Video-Based Assignments shows which MyLab Math exercises can be assigned for each video.
NEW! Note-Taking Guide (downloadable)

These printable sheets, developed by Ben Rushing (Northwestern State University) provide support
for students as they take notes in class. They include preprinted definitions, theorems, and statements
of examples but have blank space for students to write solutions to examples and sample problems.
The Note-Taking Guide corresponds to the Lecture PowerPoints that accompany the text. The Guide
can be downloaded in PDF or Word format from within MyLab Math from the Tools for Success tab.
Graphing Calculator and Excel Spreadsheet Manuals (downloadable)

Graphing Calculator Manual by Chris True, University of Nebraska
Excel Spreadsheet Manual by Stela Pudar-Hozo, Indiana University–Northwest
These manuals, both specific to this course, have been updated to support the TI-84 Plus CE (color
edition) and Excel 2016, respectively. Instructions are ordered by mathematical topic. The files can be
downloaded from within MyLab Math from the Tools for Success tab.
PREFACE 15
Student’s Solutions Manual (downloadable)

Written by John Samons (Florida State College), the Student’s Solutions Manual contains worked-out
solutions to all the odd-numbered exercises. This manual can be downloaded from within MyLab Math
within the Chapter Contents tab.
A Complete eText
Students get unlimited access to the eText within any MyLab Math course using that edition of the
textbook.
Supporting Instruction
MyLab Math comes from an experienced partner with educational expertise and an eye on the future. It
provides resources to help you assess and improve students’ results at every turn and unparalleled flex-
ibility to create a course tailored to you and your students.
Learning Catalytics™
Now included in all MyLab Math courses, this student response tool
uses students’ smartphones, tablets, or laptops to engage them in more
interactive tasks and thinking during lecture. Learning CatalyticsTM fosters
student engagement and peer-to-peer learning with real-time analytics.
Access pre-built exercises created specifically for this course.
PowerPoint® Lecture Slides (downloadable)

Classroom presentation slides feature key concepts, examples, and definitions from this text. They are
designed to be used in conjunction with the Note-Taking Guide that accompanies the text. They can be
downloaded from within MyLab Math or from Pearson’s online catalog, www.pearsonglobaleditions.com.
Learning Worksheets
Written by Salvatore Sciandra (Niagara County Community College), these worksheets include key
chapter definitions and formulas, followed by exercises for students to practice in class, for homework, or
for independent study. They are downloadable as PDFs or Word documents from within MyLab Math.
Comprehensive Gradebook
The gradebook includes enhanced report-
ing functionality such as item analysis and
a reporting dashboard to allow you to
efficiently manage your course. Student
performance data is presented at the
class, section, and program levels in an
accessible, visual manner so you’ll have
the information you need to keep your
students on track.
16 PREFACE
TestGen®
TestGen® (www.pearson.com/testgen) enables instructors to build, edit, print, and administer tests using a
computerized bank of questions developed to cover all the objectives of the text. TestGen is algorithmically
based, allowing instructors to create multiple but equivalent versions of the same question or test with the
click of a button. Instructors can also modify test bank questions or add new questions. The software and
test bank are available for download from Pearson’s online catalog, www.pearsonglobaleditions.com.
The questions are also assignable in MyLab Math.
Instructor’s Solutions Manual (downloadable)

Written by Garret J. Etgen (University of Houston) and John Samons (Florida State College), the Instruc-
tor’s Solutions Manual contains worked-out solutions to all the even-numbered exercises. It can be down-
loaded from within MyLab Math or from Pearson’s online catalog, www.pearsonglobaleditions.com.
Accessibility
Pearson works continuously to ensure our products are as accessible as possible to all students. We are
working toward achieving WCAG 2.0 Level AA and Section 508 standards, as expressed in the Pear-
son Guidelines for Accessible Educational Web Media, www.pearson.com/mylab/math/accessibility.
pearson.com/mylab/math
DIAGNOSTIC PREREQUISITE TEST 17
Diagnostic Prerequisite Test

Work all of the problems in this self-test without using a calculator. 21. u5>3u2>3 22. 19a4b-2 2 1>2
Then check your work by consulting the answers in the back of the
book. Where weaknesses show up, use the reference that follows 50 3-2
23. 2
+ -2 24. 1x1/2 + y 1/2 2 2
each answer to find the section in the text that provides the neces- 3 2
sary review.
In Problems 25–30, perform the indicated operation and write the
1. Replace each question mark with an appropriate expres- answer as a simple fraction reduced to lowest terms. All variables
sion that will illustrate the use of the indicated real number represent positive real numbers.
property:
a b a c
(A) Commutative 1 # 2: x1y + z2 = ? 25.
b
+
a
26.
bc
-
ab
(B) Associative 1 + 2: 2 + 1x + y2 = ? x2 # y 6 x x2
27. 28. ,
(C) Distributive: 12 + 32x = ? y x3 y 3 y
1 1
Problems 2–6 refer to the following polynomials: -
7 + h 7 x -1 + y -1
(A) 3x - 4 (B) x + 2 29. 30.
h x -2 - y -2
2 3
(C) 2 - 3x (D) x + 8
31. Each statement illustrates the use of one of the following real
2. Add all four.
number properties or definitions. Indicate which one.
3. Subtract the sum of (A) and (C) from the sum of (B) and (D).
Commutative 1 +, # 2 Associative 1 +, # 2 Distributive
4. Multiply (C) and (D). Identity 1 +, #2 Inverse 1 +, #2 Subtraction
5. What is the degree of each polynomial? Division Negatives Zero
6. What is the leading coefficient of each polynomial? (A) 1 - 72 - 1 - 52 = 1 - 72 + 3 - 1 - 524

In Problems 7 and 8, perform the indicated operations and simplify. (B) 5u + 13v + 22 = 13v + 22 + 5u
2
7. 5x - 3x34 - 31x - 224 (C) 15m - 22 12m + 32 = 15m - 222m + 15m - 223
8. 12x + y2 13x - 4y2 (D) 9 # 14y2 = 19 # 42y
u u
In Problems 9 and 10, factor completely. (E) =
- 1v - w2 w - v
9. x2 + 7x + 10 10. x3 - 2x2 - 15x
(F) 1x - y2 + 0 = 1x - y2
11. Write 0.35 as a fraction reduced to lowest terms.
7 32. Round to the nearest integer:
12. Write in decimal form.
8 17 5
13. Write in scientific notation: (A) (B) -
3 19
(A) 4,065,000,000,000 (B) 0.0073 33. Multiplying a number x by 4 gives the same result as sub-
tracting 4 from x. Express as an equation, and solve for x.
14. Write in standard decimal form:
34. Find the slope of the line that contains the points
(A) 2.55 * 108 (B) 4.06 * 10-4 13, - 52 and 1 - 4, 102.
15. Indicate true (T) or false (F):
35. Find the x and y coordinates of the point at which the
(A) A natural number is a rational number. graph of y = 7x - 4 intersects the x axis.
(B) A number with a repeating decimal expansion is an 36. Find the x and y coordinates of the point at which the
irrational number. graph of y = 7x - 4 intersects the y axis.
16. Give an example of an integer that is not a natural number.
In Problems 37–40, solve for x.
In Problems 17–24, simplify and write answers using positive ex- 37. x 2 = 5x
ponents only. All variables represent positive real numbers.
38. 3x 2 - 21 = 0
3 5 9u8v6
17. 61xy 2 18. 39. x 2 - x - 20 = 0
3u4v8
19. 12 * 105 2 13 * 10-3 2 20. 1x -3y 2 2 -2 40. - 6x 2 + 7x - 1 = 0
This page intentionally left blank
1 Functions
and Graphs
1.1 Functions
Introduction
1.2 Elementary Functions:
Many marine species are dependent on light from the sun. They will be found
Graphs and
Transformations near the surface of the ocean, because light intensity decreases dramatically
with depth. We use the function concept, one of the most important ideas in
1.3 Quadratic Functions mathematics, to express the precise relationship between light intensity and
1.4 Polynomial and Rational ocean depth (see Problems 63 and 64 in Section 1.5).
Functions The study of mathematics beyond the elementary level requires a firm under-
standing of a basic list of elementary functions (see the references at the back of
1.5 Exponential Functions
the book). In Chapter 1, we introduce the elementary functions and study their
1.6 Logarithmic Functions properties, graphs, and many applications.
19
20 CHAPTER 1 Functions and Graphs
1.1 Functions
■■ Equations in Two Variables We introduce the general notion of a function as a correspondence between two sets.
■■ Definition of a Function
Then we restrict attention to functions for which the two sets are both sets of real
numbers. The most useful are those functions that are specified by equations in two
■■ Functions Specified by Equations variables. We discuss the terminology and notation associated with functions, graphs
■■ Function Notation of functions, and applications.
■■ Applications
Equations in Two Variables
The graph of an equation of the form Ax + By = C, where A and B are not both
y
zero, is a line. Because a line is determined by any two of its points, such an equation
is easy to graph: Just plot any two points in its solution set and sketch the unique line
(0, 5) through them (Fig. 1).
5x 1 8y 5 40
More complicated equations in two variables, such as y = 9 - x2 or x2 = y 4,
(8, 0) are more difficult to graph. To sketch the graph of an equation, we plot enough
5
x points from its solution set in a rectangular coordinate system so that the total graph
is apparent, and then we connect these points with a smooth curve. This process is
Figure 1 called point-by-point plotting.
EXAMPLE 1 Point-by-Point Plotting Sketch the graph of each equation.

(A) y = 9 - x2 (B) x2 = y 4
SOLUTION
(A) Make up a table of solutions—that is, ordered pairs of real numbers that satisfy
the given equation. For easy mental calculation, choose integer values for x.
x -4 -3 -2 -1 0 1 2 3 4
y -7 0 5 8 9 8 5 0 -7
After plotting these solutions, if there are any portions of the graph that are
unclear, plot additional points until the shape of the graph is apparent. Then
join all the plotted points with a smooth curve (Fig. 2). Arrowheads are used
to indicate that the graph continues beyond the portion shown here with no
significant changes in shape.
y
10
(0, 9)
(21, 8) (1, 8)
(22, 5) (2, 5)
5
(23, 0) (3, 0)
x
210 25 5 10
25
(24, 27) (4, 27)
210
y 5 9 2 x2
2
Figure 2 y ∙ 9 ∙ x
(B) Again we make a table of solutions—here it may be easier to choose inte-

ger values for y and calculate values for x. Note, for example, that if y = 2,
then x = { 4; that is, the ordered pairs 14, 22 and 1 - 4, 22 are both in the
solution set.
SECTION 1.1 Functions 21
y x {9 {4 {1 0 {1 {4 {9
10 y -3 -2 -1 0 1 2 3
5
We plot these points and join them with a smooth curve (Fig. 3).
x2 5 y 4
Matched Problem 1 Sketch the graph of each equation.
x
5 10
210 25
100
(A) y = x2 - 4 (B) y 2 = 2
x + 1
25
210
Figure 3 x ∙ y4
2 Explore and Discuss 1
To graph the equation y = - x3 + 3x, we use point-by-point plotting to obtain the

graph in Figure 4.
y
x y
-1 -2 5
0 0
1 2
x
25 5
25
Figure 4
(A) Do you think this is the correct graph of the equation? Why or why not?
(B) Add points on the graph for x = - 2, - 1.5, - 0.5, 0.5, 1.5, and 2.
(C) Now, what do you think the graph looks like? Sketch your version of the graph,
adding more points as necessary.
(D) Graph this equation on a graphing calculator and compare it with your graph
from part (C).
The icon in the margin is used throughout this book to identify optional graphing
calculator activities that are intended to give you additional insight into the concepts
under discussion. You may have to consult the manual for your graphing calculator
for the details necessary to carry out these activities. For example, to graph the equa-
(A) tion in Explore and Discuss 1 on most graphing calculators, you must enter the equa-
tion (Fig. 5A) and the window variables (Fig. 5B).
As Explore and Discuss 1 illustrates, the shape of a graph may not be apparent
from your first choice of points. Using point-by-point plotting, it may be difficult
to find points in the solution set of the equation, and it may be difficult to determine
when you have found enough points to understand the shape of the graph. We will
supplement the technique of point-by-point plotting with a detailed analysis of several
basic equations, giving you the ability to sketch graphs with accuracy and confidence.
Definition of a Function
(B) Central to the concept of function is correspondence. You are familiar with corre-
Figure 5 spondences in daily life. For example,
To each person, there corresponds an annual income.

To each item in a supermarket, there corresponds a price.
To each student, there corresponds a grade-point average.
To each day, there corresponds a maximum temperature.
For the manufacture of x items, there corresponds a cost.
For the sale of x items, there corresponds a revenue.
To each square, there corresponds an area.
To each number, there corresponds its cube.
One of the most important aspects of any science is the establishment of correspon-
dences among various types of phenomena. Once a correspondence is known, pre-
dictions can be made. A cost analyst would like to predict costs for various levels of
output in a manufacturing process, a medical researcher would like to know the cor-
respondence between heart disease and obesity, a psychologist would like to predict
the level of performance after a subject has repeated a task a given number of times,
and so on.
What do all of these examples have in common? Each describes the matching of
elements from one set with the elements in a second set.
Consider Tables 1–3. Tables 1 and 2 specify functions, but Table 3 does not. Why
not? The definition of the term function will explain.
Table 1 Table 2 Table 3

Domain Range Domain Range Domain Range
Number Cube Number Square Number Square root

22 28 22 0 0
21 21 21 4 1
0 0 0 1 1 21
1 1 1 0 2
2 8 2 4 22
3
9 23
DEFINITION Function
A function is a correspondence between two sets of elements such that to each ele-
ment in the first set, there corresponds one and only one element in the second set.
The first set is called the domain, and the set of corresponding elements in the
second set is called the range.
Tables 1 and 2 specify functions since to each domain value, there corresponds
exactly one range value (for example, the cube of - 2 is - 8 and no other number). On
the other hand, Table 3 does not specify a function since to at least one domain value,
there corresponds more than one range value (for example, to the domain value 9,
there corresponds - 3 and 3, both square roots of 9).
Explore and Discuss 2

Consider the set of students enrolled in a college and the set of faculty members at
that college. Suppose we define a correspondence between the two sets by saying
that a student corresponds to a faculty member if the student is currently enrolled in
a course taught by that faculty member. Is this correspondence a function? Discuss.
Functions Specified by Equations

Most of the functions in this book will have domains and ranges that are (infinite)
sets of real numbers. The graph of such a function is the set of all points 1x, y2 in
x21 the Cartesian plane such that x is an element of the domain and y is the correspond-
ing element in the range. The correspondence between domain and range elements is
often specified by an equation in two variables. Consider, for example, the equation
x for the area of a rectangle with width 1 inch less than its length (Fig. 6). If x is the
Figure 6 length, then the area y is given by
y = x1x - 12 x Ú 1
For each input x (length), we obtain an output y (area). For example,
If x = 5, then y = 515 - 12 = 5 # 4 = 20.
If x = 1, then y = 111 - 12 = 1 # 0 = 0.
If x = 15, then y = 151 15 - 12 = 5 - 15
≈ 2.7639.
The input values are domain values, and the output values are range values. The
equation assigns each domain value x a range value y. The variable x is called an
independent variable (since values can be “independently” assigned to x from the
domain), and y is called a dependent variable (since the value of y “depends” on the
value assigned to x). In general, any variable used as a placeholder for domain val-
ues is called an independent variable; any variable that is used as a placeholder for
range values is called a dependent variable.
When does an equation specify a function?
DEFINITION Functions Specified by Equations

If in an equation in two variables, we get exactly one output (value for the depen-
dent variable) for each input (value for the independent variable), then the equation
specifies a function. The graph of such a function is just the graph of the specifying
equation.
If we get more than one output for a given input, the equation does not spec-
ify a function.
EXAMPLE 2 Functions and Equations Determine which of the following equations specify
functions with independent variable x.
(A) 4y - 3x = 8, x a real number (B) y 2 - x2 = 9, x a real number
SOLUTION
(A) Solving for the dependent variable y, we have
4y - 3x = 8
4y = 8 + 3x (1)
3
y = 2 + x
4
Since each input value x corresponds to exactly one output value 1y = 2 + 34x2,
we see that equation (1) specifies a function.
(B) Solving for the dependent variable y, we have

y 2 - x2 = 9
y 2 = 9 + x2 (2)
Reminder y = { 29 + x2
Each positive real number u Since 9 + x2 is always a positive real number for any real number x, and since
has two square roots: 2u , the each positive real number has two square roots, then to each input value x there
principal square root, and - 2u , the corresponds two output values 1y = - 29 + x2 and y = 29 + x2 2. For ex-
negative of the principal square root ample, if x = 4, then equation (2) is satisfied for y = 5 and for y = - 5. So
(see Appendix B, Section B.6). equation (2) does not specify a function.
Matched Problem 2 Determine which of the following equations specify

functions with independent variable x.
(A) y 2 - x4 = 9, x a real number (B) 3y - 2x = 3, x a real number
Since the graph of an equation is the graph of all the ordered pairs that satisfy
the equation, it is very easy to determine whether an equation specifies a function by
examining its graph. The graphs of the two equations we considered in Example 2 are
shown in Figure 7.
y y
10 10
5 5
x x
210 25 5 10 210 25 5 10
25 25
210 210
(A) 4y 2 3x 5 8 (B) y 2 2 x 2 5 9
Figure 7
In Figure 7A, notice that any vertical line will intersect the graph of the equation
4y - 3x = 8 in exactly one point. This shows that to each x value, there corresponds
exactly one y value, confirming our conclusion that this equation specifies a function.
On the other hand, Figure 7B shows that there exist vertical lines that intersect the
graph of y 2 - x2 = 9 in two points. This indicates that there exist x values to which
there correspond two different y values and verifies our conclusion that this equation
does not specify a function. These observations are generalized in Theorem 1.
THEOREM 1 Vertical-Line Test for a Function

An equation specifies a function if each vertical line in the coordinate system passes
through, at most, one point on the graph of the equation.
If any vertical line passes through two or more points on the graph of an equa-
tion, then the equation does not specify a function.
The function graphed in Figure 7A is an example of a linear function. The

vertical-line test implies that equations of the form y = mx + b, where m ∙ 0,
specify functions; they are called linear functions. Similarly, equations of the form
y = b specify functions; they are called constant functions, and their graphs are
horizontal lines. The vertical-line test implies that equations of the form x = a do not
specify functions; note that the graph of x = a is a vertical line.
In Example 2, the domains were explicitly stated along with the given equations.
In many cases, this will not be done. Unless stated to the contrary, we shall adhere
to the following convention regarding domains and ranges for functions specified by
equations:
If a function is specified by an equation and the domain is not indi-
cated, then we assume that the domain is the set of all real-number
replacements of the independent variable (inputs) that produce real
values for the dependent variable (outputs). The range is the set of all
outputs corresponding to input values.
EXAMPLE 3 Finding a Domain Find the domain of the function specified by the equation
y = 14 - x, assuming that x is the independent variable.
SOLUTION For y to be real, 4 - x must be greater than or equal to 0; that is,
4 - x Ú 0
-x Ú -4
x … 4 Sense of inequality reverses when both sides are divided by - 1.
Domain: x … 4 (inequality notation) or 1- ∞, 44 (interval notation)
Matched Problem 3 Find the domain of the function specified by the equa-
tion y = 1x - 2, assuming x is the independent variable.
Function Notation
We have seen that a function involves two sets, a domain and a range, and a cor-
respondence that assigns to each element in the domain exactly one element in the
range. Just as we use letters as names for numbers, now we will use letters as names
for functions. For example, f and g may be used to name the functions specified by
the equations y = 2x + 1 and y = x2 + 2x - 3:
f : y = 2x + 1
g: y = x2 + 2x - 3 (3)
If x represents an element in the domain of a function f, then we frequently use
f the symbol
f 1x2
x f ( x)
in place of y to designate the number in the range of the function f to which x is
paired (Fig. 8). This symbol does not represent the product of f and x. The symbol
DOMAIN RANGE f1x2 is read as “f of x,” “f at x,” or “the value of f at x.” Whenever we write y = f1x2,
Figure 8 we assume that the variable x is an independent variable and that both y and f1x2 are
dependent variables.
Using function notation, we can now write functions f and g in equation (3) as
f1x2 = 2x + 1 and g1x2 = x2 + 2x - 3
Let us find f132 and g1 - 52. To find f132, we replace x with 3 wherever x occurs in
f1x2 = 2x + 1 and evaluate the right side:
f1x2 = 2x + 1
f132 = 2 # 3 + 1
= 6 + 1 = 7 For input 3, the output is 7.
Therefore,
f132 = 7 The function f assigns the range value 7 to the domain value 3.
To find g1 - 52, we replace each x by - 5 in g1x2 = x2 + 2x - 3 and evaluate
the right side:
g1x2 = x2 + 2x - 3
g1 - 52 = 1 - 52 2 + 21 - 52 - 3
= 25 - 10 - 3 = 12 For input - 5, the output is 12.
Therefore,
g1 - 52 = 12 The function g assigns the range value 12 to the domain value - 5.
It is very important to understand and remember the definition of f1x2:
For any element x in the domain of the function f, the symbol f 1x 2 rep-
resents the element in the range of f corresponding to x in the domain
of f. If x is an input value, then f 1 x 2 is the corresponding output value.
If x is an element that is not in the domain of f, then f is not defined at x
and f 1 x 2 does not exist.
EXAMPLE 4 Function Evaluation For f1x2 = 12> 1x - 22, g1x2 = 1 - x2, and
h1x2 = 1x - 1, evaluate:
(A) f162 (B) g1 - 22 (C) h1 - 22 (D) f102 + g112 - h1102
SOLUTION
12 12
(A) f162 = = = 3
6 - 2 4
(B) g1 - 22 = 1 - 1 - 22 2 = 1 - 4 = - 3
(C) h1 - 22 = 1 - 2 - 1 = 1 - 3
Reminder
Dashed boxes are used throughout But 1 - 3 is not a real number. Since we have agreed to restrict the domain of
the book to represent steps that are a function to values of x that produce real values for the function, - 2 is not in
usually performed mentally. the domain of h, and h1 - 22 does not exist.
12
(D) f102 + g112 - h1102 = + 11 - 12 2 - 110 - 1
0 - 2
12
= + 0 - 19
-2
= -6 - 3 = -9
Matched Problem 4 Use the functions in Example 4 to find
f132
(A) f1 - 22 (B) g1 - 12 (C) h1 - 82 (D)
h152
EXAMPLE 5 Finding Domains Find the domains of functions f, g, and h:

12
f1x2 = g1x2 = 1 - x2 h1x2 = 2x - 1
x - 2
SOLUTION Domain of f: 12> 1x - 22 represents a real number for all replacements
of x by real numbers except for x = 2 (division by 0 is not defined). Thus, f122
does not exist, and the domain of f is the set of all real numbers except 2. We often
indicate this by writing
12
f1x2 = x ∙ 2
x - 2
Domain of g: The domain is R, the set of all real numbers, since 1 - x2 represents
a real number for all replacements of x by real numbers.
Domain of h: The domain is the set of all real numbers x such that 2x - 1 is a real
number; so
x - 1 Ú 0
x Ú 1 or, in interval notation, 31, ∞ 2
Matched Problem 5 Find the domains of functions F, G, and H:

5
F1x2 = x2 - 3x + 1 G1x2 = H1x2 = 22 - x
x + 3
In addition to evaluating functions at specific numbers, it is important to be able

to evaluate functions at expressions that involve one or more variables. For example,
the difference quotient
f1x + h2 - f1x2
x and x + h in the domain of f, h ∙ 0
h
is studied extensively in calculus.
CONCEPTUAL INSIGHT
In algebra, you learned to use parentheses for grouping variables. For example,
21x + h2 = 2x + 2h
Now we are using parentheses in the function symbol f1 x 2. For example, if
f1 x2 = x 2, then
f1x + h2 = 1x + h2 2 = x2 + 2xh + h2
Note that f1x2 + f1h2 = x2 + h2 ∙ f1x + h2. That is, the function name
f does not distribute across the grouped variables 1x + h2, as the “2” does in
21x + h2 (see Appendix B, Section B.2).
EXAMPLE 6 Using Function Notation For f1x2 = x2 - 2x + 7, find

(A) f1a2 (B) f1a + h2
f1a + h2 - f1a2
(C) f1a + h2 - f1a2 (D) , h ∙ 0
h
SOLUTION
(A) f1a2 = a2 - 2a + 7
(B) f1a + h2 = 1a + h2 2 - 21a + h2 + 7 = a2 + 2ah + h2 - 2a - 2h + 7
(C) f1a + h2 - f1a2 = 1a2 + 2ah + h2 - 2a - 2h + 72 - 1a2 - 2a + 72
= 2ah + h2 - 2h
f1a + h2 - f1a2 2ah + h2 - 2h h12a + h - 22 h
(D) = = Because h ∙ 0, = 1.
h h h h
= 2a + h - 2
Matched Problem 6 Repeat Example 6 for f1x2 = x2 - 4x + 9.
Applications
If we reduce the price of a product, will we generate more revenue? If we increase
production, will our profits rise? Profit–loss analysis is a method for answering such
questions in order to make sound business decisions.
Here are the basic concepts of profit–loss analysis: A manufacturing company
has costs, C, which include fixed costs such as plant overhead, product design, setup,
and promotion and variable costs that depend on the number of items produced. The
revenue, R, is the amount of money received from the sale of its product. The com-
pany takes a loss if R 6 C, breaks even if R = C, and has a profit if R 7 C. The
profit P is equal to revenue minus cost; that is, P = R - C. (So the company takes
a loss if P 6 0, breaks even if P = 0, and has a profit if P 7 0.) To predict its rev-
enue, a company uses a price–demand function, p1x2, determined using historical
data or sampling techniques, that specifies the relationship between the demand x and
the price p. A point (x, p) is on the graph of the price–demand function if x items can
be sold at a price of $p per item. (Normally, a reduction in the price p will increase
the demand x, so the graph of the price–demand function is expected to go downhill
as you move from left to right.) The revenue R is equal to the number of items sold
multiplied by the price per item; that is, R = xp.
Cost, revenue, and profit can be written as functions C(x), R(x), and P(x) of the
independent variable x, the number of items manufactured and sold. The functions
C(x), R(x), P(x), and p(x) often have the following forms, where a, b, m, and n are
positive constants determined from the context of a particular problem:
Cost function
C1x2 = a + bx C = fixed costs + variable costs
Price–demand function
p1x2 = m - nx x is the number of items that can be sold at $p per item
Revenue function
R1x2 = xp R = number of items sold * price per item
= x1m - nx2
Profit function
P1x2 = R1x2 - C1x2
= x1m - nx2 - 1a + bx2
Do not confuse the price–demand function p(x) with the prof-
! CAUTION it function P(x). Price is always denoted by the lowercase p.
Profit is always denoted by the uppercase P. Note that the revenue and profit func-
tions, R(x) and P(x), depend on the price–demand function p(x), but C(x) does not.
Example 7 and Matched Problem 7 provide an introduction to profit–loss analysis.
EXAMPLE 7 Price–Demand and Revenue A manufacturer of a popular digital camera whole-

sales the camera to retail outlets throughout the United Kingdom. Using statistical
methods, the financial department in the company produced the price–demand data
in Table 4, where p is the wholesale price per camera at which x million cameras are
sold. Notice that as the price goes down, the number sold goes up.
Table 4 Price–Demand
x (millions) p 1$2
2 87
5 68
8 53
12 37
Using special analytical techniques (regression analysis), an analyst obtained

Table 5 Revenue the following price–demand function to model the Table 4 data:
x (millions) R 1 x2 (million $) p1x2 = 94.8 - 5x 1 … x … 15 (4)
1 90
3
(A) Plot the data in Table 4. Then sketch a graph of the price–demand function in
6 the same coordinate system.
9 (B) What is the company’s revenue function for this camera, and what is its domain?
12 (C) Complete Table 5, computing revenues to the nearest million dollars.
15
(D) Plot the data in Table 5. Then sketch a graph of the revenue function using these
points.
(E) Graph the revenue function on a graphing calculator.
SOLUTION
(A) The four data points are plotted in Figure 9. Note that p112 = 89.8 and
p1152 = 19.8. So the graph of the price–demand function is the line through
(1, 89.8) and (15, 19.8) (see Fig. 9).
p(x)
100
Price per camera ($)
50
x
5 10 15
Million cameras
Figure 9 Price–demand
In Figure 9, notice that the model approximates the actual data in Table 4, and it is
assumed that it gives realistic and useful results for all other values of x between
1 million and 15 million.
(B) R1x2 = x p1x2 = x194.8 - 5x2 million dollars
Domain: 1 … x … 15
[Same domain as the price–demand function, equation (4).]
R(x) (C) Table 5 Revenue

500 x (millions) R 1 x2 (million $)
1 90
400
3 239
Million dollars
6 389
300
9 448
200 12 418
15 297
100
(D) The six points from Table 5 are plotted in Figure 10. The graph of the revenue
x
5 10 15 function is the smooth curve drawn through those six points.
Million cameras
(E) Figure 11 shows the graph of R1x2 = x194.8 - 5x2 on a graphing calculator.
Figure 10
Matched Problem 7 The financial department in Example 7, using statistical
techniques, produced the data in Table 6, where C1x2 is the cost in millions of dollars
500
for manufacturing and selling x million cameras.
Table 6 Cost Data

x (millions) C 1 x2 1million $ 2
1 175
1 15
0 5 260
8 305
Figure 11
12 395
Using special analytical techniques (regression analysis), an analyst produced the

following cost function to model the Table 6 data:
C1x2 = 156 + 19.7x 1 … x … 15 (5)
(A) Plot the data in Table 6. Then sketch a graph of equation ( 5) in the same coor-
dinate system.
(B) Using the revenue function from Example 7(B), what is the company’s profit
function for this camera, and what is its domain?
(C) Complete Table 7, computing profits to the nearest million dollars.
Table 7 Profit
x (millions) P 1x2 (million $)
1 -86
3
6
9
12
15
(D) Plot the data in Table 7. Then sketch a graph of the profit function using these
points.
(E) Graph the profit function on a graphing calculator.
Exercises 1.1
A In Problems 1–8, use point-by-point plotting to sketch the graph 19. y 20. y
of each equation. 10 10
1. y = x + 1 2. x = y + 1
5 5
3. x = y 2 4. y = x2
x x
5. y = x3 6. x = y 3 210 25 5 10 210 25 5 10
7. xy = - 6 8. xy = 12 25 25
210 210
Indicate whether each table in Problems 9–14 specifies a function.
9. Domain Range 10. Domain Range
In Problems 21–28, each equation specifies a function with
3 0 21 5 independent variable x. Determine whether the function is linear,
5 1 22 7 constant, or neither.
7 2 23 9 1 1
21. y = - 3x + 22. y = 4x +
8 x
11. Domain Range 12. Domain Range 23. 7x + 5y = 3 24. 2x - 4y - 6 = 0
3 5 8 0
25. y - 5x = 4 - 3x2 26. x + xy + 1 = 0
6 9 1 y - x 3 + 2x
4 7 2 27. y - x2 + 2 = 10 - x2 + 28.= 1
5 8 10 3
2 4
In Problems 29–36, use point-by-point plotting to sketch the graph
13. Domain Range 14. Domain Range of each function.
x
3 22 29. f1x2 = 1 - x 30. f1x2 = - 3
5 2
6 21
6
9 0 31. f1x2 = x2 - 1 32. f1x2 = 3 - x2
6
12 1
33. f1x2 = 4 - x3 34. f1x2 = x3 - 2
8 -6
Indicate whether each graph in Problems 15–20 specifies a 35. f1x2 = 36. f1x2 =
x x
function.
15. y 16. y In Problems 37 and 38, the three points in the table are on the
10 10
graph of the indicated function f. Do these three points provide
sufficient information for you to sketch the graph of y = f1x2? Add
5 5 more points to the table until you are satisfied that your sketch is a
good representation of the graph of y = f1x2 for - 5 … x … 5.
x x
210 25 5 10 210 25 5 10 37. x -1 0 1 2x
f1x2 =
25 25
f1x2 -1 0 1 x2 + 1
210 210 38. x 0 1 2 3x2
f1x2 =
17. y 18. y f1x2 0 1 2 x2 + 2
10 10
In Problems 39–46, use the following graph of a function f to de-
5 5
termine x or y to the nearest integer, as indicated. Some problems
may have more than one answer.
x x f (x) y 5 f (x)
210 25 5 10 210 25 5 10
10
25 25
210 210 5
x
210 25 5 10
25
210
39. y = f1 - 52 40. y = f142 81. The area of a rectangle is 25 sq in. Express the perimeter
P1w2 as a function of the width w, and state the domain of
41. y = f132 42. y = f1 - 22
this function.
43. f1x2 = 0, x 6 0 44. f1x2 = - 2
82. The area of a rectangle is 216 sq ft. Express the perimeter
45. f1x2 = - 5 46. f1x2 = 0 P1l2 as a function of the length l, and state the domain of this
function.
B In Problems 47–52, find the domain of each function.
83. The perimeter of a rectangle is 100 m. Express the area A1l2
47. F1x2 = 2x3 - x2 + 3 48. H1x2 = 7 - 2x2 - x4 as a function of the length l, and state the domain of this
x - 2 x + 5 function.
49. f1x2 = 50. g1x2 =
x + 4 x - 6 84. The perimeter of a rectangle is 160 m. Express the area
8 A1w2 as a function of the width w, and state the domain of
51. g1x2 = 27 - x 52. F1x2 =
29 + x this function.
In Problems 53–60, does the equation specify a function with

independent variable x? If so, find the domain of the function. If
not, find a value of x to which there corresponds more than one
value of y.
Applications
53. 2x + 5y = 10 54. 6x - 7y = 21 85. Price–demand. A company manufactures memory chips for
microcomputers. Its marketing research department, using
55. y1x + y2 = 4 56. x1x + y2 = 4
statistical techniques, collected the data shown in Table 8,
-3 3
57. x + y = 27 58. x2 + y 2 = 9 where p is the wholesale price per chip at which x million
59. x3 - y 2 = 0 60. 1x - y 3 = 0 chips can be sold. Using special analytical techniques (re-
gression analysis), an analyst produced the following price–
In Problems 61–74, find and simplify the expression if demand function to model the data:
f1x2 = x2 - 4. p1x2 = 75 - 3x 1 … x … 20
61. f15x2 62. f1 - 3x2
63. f1x + 32 64. f1x - 12
x (millions) p1$2
65. f1x2 2 66. f1x3 2 1 72
4
67. f1 1x2 68. f1 2 x2 4 63
9 48
69. f122 + f1h2 70. f1 - 32 + f1h2
14 33
71. f12 + h2 72. f1 - 3 + h2 20 15
73. f12 + h2 - f122 74. f1 - 3 + h2 - f1 - 32 (A) Plot the data points in Table 8, and sketch a graph of the
price–demand function in the same coordinate system.
C In Problems 75–80, find and simplify each of the following,
assuming h ∙ 0 in (C). (B) What would be the estimated price per chip for a de-
mand of 7 million chips? For a demand of 11 million
(A) f1x + h2
chips?
(B) f1x + h2 - f1x2
86. Price–demand. A company manufactures notebook computers.
f1x + h2 - f1x2 Its marketing research department, using statistical tech-
(C) niques, collected the data shown in Table 9, where p is the
h
wholesale price per computer at which x thousand computers
75. f1x2 = 4x - 3 76. f1x2 = - 3x + 9 can be sold. Using special analytical techniques (regression
2 analysis), an analyst produced the following price–demand
77. f1x2 = 4x - 7x + 6 78. f1x2 = 3x2 + 5x - 8 function to model the data:
79. f1x2 = x120 - x2 80. f1x2 = x1x + 502 p1x2 = 2,000 - 60x 1 … x … 25
Problems 81–84 refer to the area A and perimeter P of a rectangle
with length l and width w (see the figure). Table 9 Price–Demand
x (thousands) p1$2
A 5 lw 1 1,940
w 8 1,520
P 5 2l 1 2w
16 1,040
21 740
l 25 500
(A) Plot the data points in Table 9, and sketch a graph of the (B) Complete Table 12, computing profits to the nearest mil-
price–demand function in the same coordinate system. lion dollars.
(B) What would be the estimated price per computer for a
Table 12 Profit
demand of 11,000 computers? For a demand of 18,000
computers? x (millions) P 1x2 (million $)
87. Revenue. 1 -69
4
(A) Using the price–demand function
8
p1x2 = 75 - 3x 1 … x … 20 12
from Problem 85, write the company’s revenue function 16
and indicate its domain. 20
(B) Complete Table 10, computing revenues to the nearest
million dollars. (C) Plot the points in part (B) and sketch a graph of the
profit function using these points.
Table 10 Revenue
90. Profit. The financial department for the company in Problems
x (millions) R 1 x2 (million $) 86 and 88 established the following cost function for producing
1 72 and selling x thousand notebook computers:
4 C1x2 = 4,000 + 500x thousand dollars
8
12 (A) Write a profit function for producing and selling x thousand
16
notebook computers and indicate its domain.
20 (B) Complete Table 13, computing profits to the nearest
thousand dollars.
(C) Plot the points from part (B) and sketch a graph of the
revenue function using these points. Choose millions for Table 13 Profit
the units on the horizontal and vertical axes.
x (thousands) P 1x2 (thousand $)
88. Revenue. 1 -2,560
(A) Using the price–demand function 5
p1x2 = 2,000 - 60x 1 … x … 25 10
15
from Problem 86, write the company’s revenue function 20
and indicate its domain. 25
(B) Complete Table 11, computing revenues to the nearest
thousand dollars. (C) Plot the points in part (B) and sketch a graph of the
profit function using these points.
Table 11 Revenue
91. Muscle contraction. In a study of the speed of muscle
x (thousands) R 1 x2 (thousand $) contraction in frogs under various loads, British biophysicist
1 1,940 A. W. Hill determined that the weight w (in grams) placed on
5 the muscle and the speed of contraction v (in centimeters per
10 second) are approximately related by an equation of the form
15 1w + a2 1v + b2 = c
20
where a, b, and c are constants. Suppose that for a certain
25
muscle, a = 15, b = 1, and c = 90. Express v as a func-
tion of w. Find the speed of contraction if a weight of 16 g is
(C) Plot the points from part (B) and sketch a graph of the placed on the muscle.
revenue function using these points. Choose thousands
92. Politics. The percentage s of seats in the House of Commons
for the units on the horizontal and vertical axes.
won by Conservatives and the percentage v of votes cast for
89. Profit. The financial department for the company in Conservatives (when expressed as decimal fractions) are
Problems 85 and 87 established the following cost function related by the equation
for producing and selling x million memory chips:
8v - 5s = 1 0 6 s 6 1, 0.125 6 v 6 0.75
C1x2 = 125 + 16x million dollars
(A) Express v as a function of s and find the percentage of votes
(A) Write a profit function for producing and selling x required for the Conservatives to win 51% of the seats.
million memory chips and indicate its domain.
(B) Express s as a function of v and find the percentage of
seats won if the Conservatives receive 51% of the votes.
Answers to Matched Problems (B) P1x2 = R1x2 - C1x2

1. (A) (B) = x194.8 - 5x2 - 1156 + 19.7x2;
y y 5 x2 2 4 y 100
y2 5 domain: 1 … x … 15
x2 1 1
15 0
10
(C) Table 7 Profit
10 5 x (millions) P 1x2 (million $)
1 -86
5 x
25 5 3 24
x 25 6 115
210 25 5 10
9 115
25 0
210 12 25
15 -155
2. (A) Does not specify a function
(B) Specifies a function (D) P(x)
200
3. x Ú 2 (inequality notation) or 32, ∞ 2 (interval notation)
4. (A) - 3 (B) 0 100
Million dollars
(C) Does not exist (D) 6
x
5. Domain of F: R; domain of G: all real numbers except - 3; 0 5 10 15
domain of H: x … 2 (inequality notation) or 1 - ∞ , 24 2100

(interval notation)
2200
6. (A) a2 - 4a + 9 Million cameras
(B) a2 + 2ah + h2 - 4a - 4h + 9
(C) 2ah + h2 - 4h (E) 200
(D) 2a + h - 4
7. (A) C(x) 1 15
500
400
Million dollars
2200
300
200
100
x
0 5 10 15
Million cameras
1.2 Elementary Functions: Graphs and Transformations

■■ A Beginning Library of Elementary Each of the functions
Functions
g1x2 = x2 - 4 h1x2 = 1x - 42 2 k1x2 = - 4x2
■■ Vertical and Horizontal Shifts
can be expressed in terms of the function f1x2 = x2:
■■ Reflections, Stretches, and Shrinks
g1x2 = f1x2 - 4 h1x2 = f1x - 42 k1x2 = - 4f1x2
■■ Piecewise-Defined Functions
In this section, we will see that the graphs of functions g, h, and k are closely related
to the graph of function f. Insight gained by understanding these relationships will
help us analyze and interpret the graphs of many different functions.
A Beginning Library of Elementary Functions

As you progress through this book, you will repeatedly encounter a small number
of elementary functions. We will identify these functions, study their basic prop-
erties, and include them in a library of elementary functions (see the references
SECTION 1.2 Elementary Functions: Graphs and Transformations 35
at the back of the book). This library will become an important addition to your mathe-
matical toolbox and can be used in any course or activity where mathematics is applied.
We begin by placing six basic functions in our library.
DEFINITION Basic Elementary Functions

f1x2 = x Identity function
h1x2 = x2 Square function
m1x2 = x3 Cube function
n1x2 = 1x Square root function
3
p1x2 = 2x Cube root function
g1x2 = ∙x∙ Absolute value function
These elementary functions can be evaluated by hand for certain values of x and
with a calculator for all values of x for which they are defined.
EXAMPLE 1 Evaluating Basic Elementary Functions Evaluate each basic elementary function at
(A) x = 64 (B) x = - 12.75
Round any approximate values to four decimal places.
SOLUTION
(A) f1642 = 64
h1642 = 642 = 4,096 Use a calculator.
3
m1642 = 64 = 262,144 Use a calculator.
n1642 = 264 = 8
3
p1642 = 2 64 = 4
g1642 = ∙ 64 ∙ = 64
(B) f1 - 12.752 = - 12.75
h1 - 12.752 = 1 - 12.752 2 = 162.5625 Use a calculator.
m1 - 12.752 = 1 - 12.752 3 ≈ - 2,072.6719 Use a calculator.
n1 - 12.752 = 2- 12.75 Not a real number.
3
p1 - 12.752 = 2 - 12.75 ≈ - 2.3362 Use a calculator.
g1 - 12.752 = ∙ - 12.75 ∙ = 12.75
Matched Problem 1 Evaluate each basic elementary function at
(A) x = 729 (B) x = - 5.25
Round any approximate values to four decimal places.
Remark Most computers and graphing calculators use ABS(x) to represent the abso-
lute value function. The following representation can also be useful:
∙ x ∙ = 2x2
Figure 1 shows the graph, range, and domain of each of the basic elementary functions.
f (x) h(x) m(x)

5 5 5
x x x
25 5 25 5 25 5
25 25 25
(A) Identity function (B) Square function (C) Cube function

f (x) 5 x h(x) 5 x 2 m(x) 5 x3
Domain: R Domain: R Domain: R
Reminder Range: R Range: [0, `) Range: R
Letters used to designate these
functions may vary from context n(x) p(x) g(x)
to context; R is the set of all real 5 5 5
numbers.
x x x
25 5 25 5 25 5
25 25 25
(D) Square root function (E) Cube root function (F) Absolute value function
n(x) 5 Ïx p(x) 5 Ïx
3
g(x) 5 ) x)
Domain: [0, `) Domain: R Domain: R
Range: [0, `) Range: R Range: [0, `)
Figure 1 Some basic functions and their graphs
CONCEPTUAL INSIGHT
Absolute Value In beginning algebra, absolute value is often interpreted as dis-
tance from the origin on a real number line (see Appendix B, Section B.1).
distance 5 6 5 2(26) distance 5 5
210 25 0 5 10
If x 6 0, then - x is the positive distance from the origin to x, and if x 7 0, then

x is the positive distance from the origin to x. Thus,
- x if x 6 0
∙ x∙ = e
x if x Ú 0
Vertical and Horizontal Shifts

If a new function is formed by performing an operation on a given function, then
the graph of the new function is called a transformation of the graph of the original
function. For example, graphs of y = f1x2 + k and y = f1x + h2 are transforma-
tions of the graph of y = f1x2.
Let f1x2 = x2.

(A) Graph y = f1x2 + k for k = - 4, 0, and 2 simultaneously in the same coordi-
nate system. Describe the relationship between the graph of y = f1x2 and the
graph of y = f1x2 + k for any real number k.
(B) Graph y = f1x + h2 for h = - 4, 0, and 2 simultaneously in the same coordi-

nate system. Describe the relationship between the graph of y = f1x2 and the
graph of y = f1x + h2 for any real number h.
EXAMPLE 2 Vertical and Horizontal Shifts

(A) How are the graphs of y = ∙ x ∙ + 4 and y = ∙ x ∙ - 5 related to the graph of
y = ∙ x ∙ ? Confirm your answer by graphing all three functions simultaneously
in the same coordinate system.
(B) How are the graphs of y = ∙ x + 4 ∙ and y = ∙ x - 5 ∙ related to the graph of
SOLUTION
(A) The graph of y = ∙ x ∙ + 4 is the same as the graph of y = ∙ x ∙ shifted upward
4 units, and the graph of y = ∙ x ∙ - 5 is the same as the graph of y = ∙ x ∙
shifted downward 5 units. Figure 2 confirms these conclusions. [It appears that
the graph of y = f1x2 + k is the graph of y = f1x2 shifted up if k is positive
and down if k is negative.]
(B) The graph of y = ∙ x + 4 ∙ is the same as the graph of y = ∙ x ∙ shifted to the
left 4 units, and the graph of y = ∙ x - 5 ∙ is the same as the graph of y = ∙ x ∙
shifted to the right 5 units. Figure 3 confirms these conclusions. [It appears that
the graph of y = f1x + h2 is the graph of y = f1x2 shifted right if h is nega-
tive and left if h is positive—the opposite of what you might expect.]
y y
y 5 |x| 1 4 y 5 |x 1 4|
y 5 |x| y 5 |x|
10 10
y 5 |x 2 5|
5 y 5 |x| 2 5 5
x x
210 25 5 10 210 25 5 10
25 25
210 210
Figure 2 Vertical shifts Figure 3 Horizontal shifts
Matched Problem 2
(A) How are the graphs of y = 1x + 5 and y = 1x - 4 related to the graph of
y = 1x ? Confirm your answer by graphing all three functions simultaneously
(B) How are the graphs of y = 1x + 5 and y = 1x - 4 related to the graph of
y = 1x? Confirm your answer by graphing all three functions simultaneously
Comparing the graphs of y = f1x2 + k with the graph of y = f1x2, we see that
the graph of y = f1x2 + k can be obtained from the graph of y = f1x2 by vertically
translating (shifting) the graph of the latter upward k units if k is positive and down-
ward ∙ k ∙ units if k is negative. Comparing the graphs of y = f1x + h2 with the graph
of y = f1x2, we see that the graph of y = f1x + h2 can be obtained from the graph
of y = f1x2 by horizontally translating (shifting) the graph of the latter h units to
the left if h is positive and ∙ h ∙ units to the right if h is negative.
EXAMPLE 3 Vertical and Horizontal Translations (Shifts) The graphs in Figure 4 are either
horizontal or vertical shifts of the graph of f1x2 = x2. Write appropriate equations
for functions H, G, M, and N in terms of f.
y H f G M y f N
5 5
x x
25 5 25 5
25 25
(A) (B)
Figure 4 Vertical and horizontal shifts
SOLUTION Functions H and G are vertical shifts given by

H1x2 = x2 + 2 G1x2 = x2 - 4
Functions M and N are horizontal shifts given by
M1x2 = 1x + 22 2 N1x2 = 1x - 32 2
Matched Problem 3 The graphs in Figure 5 are either horizontal or vertical

3
shifts of the graph of f1x2 = 2 x. Write appropriate equations for functions H, G,
M, and N in terms of f.
y y
5 5
M
H
f f
G
x x
25 5 25 5
N
25 25
(A) (B)
Figure 5 Vertical and horizontal shifts
Reflections, Stretches, and Shrinks

We now investigate how the graph of y = Af1x2 is related to the graph of y = f1x2
for different real numbers A.
(A) Graph y = Ax2 for A = 1, 4, and 14 simultaneously in the same coordinate

system.
(B) Graph y = Ax2 for A = - 1, - 4, and - 14 simultaneously in the same coordinate
system.
(C) Describe the relationship between the graph of h1x2 = x2 and the graph of
G1x2 = Ax2 for any real number A.
Comparing y = Af1x2 to y = f1x2, we see that the graph of y = Af1x2 can be

obtained from the graph of y = f1x2 by multiplying each ordinate value of the latter
by A. The result is a vertical stretch of the graph of y = f1x2 if A 7 1, a vertical
shrink of the graph of y = f1x2 if 0 6 A 6 1, and a reflection in the x axis if
A = - 1. If A is a negative number other than - 1, then the result is a combination of
a reflection in the x axis and either a vertical stretch or a vertical shrink.
EXAMPLE 4 Reflections, Stretches, and Shrinks

(A) How are the graphs of y = 2 ∙ x ∙ and y = 0.5 ∙ x ∙ related to the graph of
(B) How is the graph of y = - 2 ∙ x ∙ related to the graph of y = ∙ x ∙ ? Confirm
your answer by graphing both functions simultaneously in the same coordinate
system.
SOLUTION
(A) The graph of y = 2 ∙ x ∙ is a vertical stretch of the graph of y = ∙ x ∙ by a factor
of 2, and the graph of y = 0.5 ∙ x ∙ is a vertical shrink of the graph of y = ∙ x ∙
by a factor of 0.5. Figure 6 confirms this conclusion.
(B) The graph of y = - 2 ∙ x ∙ is a reflection in the x axis and a vertical stretch of the
graph of y = ∙ x ∙ . Figure 7 confirms this conclusion.
y
y y 5 2| x| y 5 |x|
y 5 |x| 10
10
5 y 5 0.5| x | 5
x
x 210 25 5 10
210 25 5 10
25
25
210
210 y 5 22| x |
Figure 6 Vertical stretch and shrink Figure 7 Reflection and vertical stretch
Matched Problem 4
(A) How are the graphs of y = 2x and y = 0.5x related to the graph of y = x?
Confirm your answer by graphing all three functions simultaneously in the
same coordinate system.
(B) How is the graph of y = - 0.5x related to the graph of y = x? Confirm your
answer by graphing both functions in the same coordinate system.
The various transformations considered above are summarized in the following

box for easy reference:
SUMMARY Graph Transformations

Vertical Translation:
k 7 0 Shift graph of y = f1x2 up k units.
y = f1x2 + k e
k 6 0 Shift graph of y = f1x2 down ∙ k ∙ units.
Horizontal Translation:
h 7 0 Shift graph of y = f1x2 left h units.
y = f1x + h2 e
h 6 0 Shift graph of y = f1x2 right ∙ h ∙ units.
Reflection:
y = - f1x2 Reflect the graph of y = f1x2 in the x axis.
Vertical Stretch and Shrink:
A 7 1 Stretch graph of y = f1x2 vertically

by multiplying each ordinate value by A.
y = Af1x2 µ
0 6 A 6 1 Shrink graph of y = f1x2 vertically
by multiplying each ordinate value by A.

Explain why applying any of the graph transformations in the summary box to a
linear function produces another linear function.
EXAMPLE 5 Combining Graph Transformations Discuss the relationship between the graphs
of y = - ∙ x - 3 ∙ + 1 and y = ∙ x ∙ . Confirm your answer by graphing both func-
tions simultaneously in the same coordinate system.
SOLUTION The graph of y = - ∙ x - 3 ∙ + 1 is a reflection of the graph of y = ∙ x ∙
in the x axis, followed by a horizontal translation of 3 units to the right and a vertical
translation of 1 unit upward. Figure 8 confirms this description.
y
y 5 |x|
5
x
25 5
y
G y 5 x3
5
25
y 5 2| x 2 3 | 1 1
25 5
x Figure 8 Combined transformations
Matched Problem 5 The graph of y = G1x2 in Figure 9 involves a reflection

and a translation of the graph of y = x3. Describe how the graph of function G is
Figure 9 Combined transformations related to the graph of y = x3 and find an equation of the function G.
Piecewise-Defined Functions
Earlier we noted that the absolute value of a real number x can be defined as
-x if x 6 0
∙x∙ = e
x if x Ú 0
Notice that this function is defined by different rules for different parts of its domain.
Functions whose definitions involve more than one rule are called piecewise-defined
functions. Graphing one of these functions involves graphing each rule over the ap-
propriate portion of the domain (Fig. 10). In Figure 10C, notice that an open dot is
used to show that the point 10, - 22 is not part of the graph and a solid dot is used to
show that (0, 2) is part of the graph.
y y y
5 5 5
x x x
25 5 25 5 25 5
25 25 25
(A) y 5 x 2 2 2 (B) y 5 2 2 x 2 x 2 2 2 if x , 0
(C) y 5
x,0 x$0 2 2 x 2 if x $ 0
Figure 10 Graphing a piecewise-defined function
EXAMPLE 6 Graphing Piecewise-Defined Functions Graph the piecewise-defined function
x + 1 if 0 … x 6 2
g1x2 = e
0.5x if x Ú 2
SOLUTION If 0 … x 6 2, then the first rule applies and the graph of g lies on
the line y = x + 1 (a vertical shift of the identity function y = x). If x = 0, then
(0, 1) lies on y = x + 1; we plot (0, 1) with a solid dot (Fig. 11) because g102 = 1.
If x = 2, then (2, 3) lies on y = x + 1; we plot (2, 3) with an open dot because
g122 ∙ 3. The line segment from (0, 1) to (2, 3) is the graph of g for 0 … x 6 2.
If x Ú 2, then the second rule applies and the graph of g lies on the line y = 0.5x
(a vertical shrink of the identity function y = x). If x = 2, then (2, 1) lies on the
line y = 0.5x; we plot (2, 1) with a solid dot because g122 = 1. The portion of
y = 0.5x that starts at (2, 1) and extends to the right is the graph of g for x Ú 2.
g(x)
x
1 2 3 4 5
Figure 11
Matched Problem 6 Graph the piecewise-defined function
- 2x + 4 if 0 … x … 2
h1x2 = e
x - 1 if x 7 2
As the next example illustrates, piecewise-defined functions occur naturally in

many applications.
EXAMPLE 7 Natural Gas Rates Easton Utilities uses the rates shown in Table 1 to compute
the monthly cost of natural gas for each customer. Write a piecewise definition
for the cost of consuming x CCF (cubic hundred feet) of natural gas and graph the
function.
Table 1 Charges per Month
$0.7866 per CCF for the first 5 CCF
$0.4601 per CCF for the next 35 CCF
$0.2508 per CCF for all over 40 CCF
SOLUTION If C1x2 is the cost, in dollars, of using x CCF of natural gas in one
month, then the first line of Table 1 implies that
C1x2 = 0.7866x if 0 … x … 5
Note that C152 = 3.933 is the cost of 5 CCF. If 5 6 x … 40, then x - 5 repre-
sents the amount of gas that cost $0.4601 per CCF, 0.46011x - 52 represents the
cost of this gas, and the total cost is
C1x2 = 3.933 + 0.46011x - 52
If x 7 40, then
C(x)
C1x2 = 20.0365 + 0.25081x - 402
$30
where 20.0365 = C1402, the cost of the first 40 CCF. Combining all these equa-
tions, we have the following piecewise definition for C1x2:
$20
(40, 20.0365)
0.7866x if 0 … x … 5
$10 C1x2 = c 3.933 + 0.46011x - 52 if 5 6 x … 40
20.0365 + 0.25081x - 402 if x 7 40
(5, 3.933)
x
10 20 30 40 50 60 To graph C, first note that each rule in the definition of C represents a transforma-
Figure 12 Cost of purchasing x CCF tion of the identity function f1x2 = x. Graphing each transformation over the indi-
of natural gas cated interval produces the graph of C shown in Figure 12.
Matched Problem 7 Trussville Utilities uses the rates shown in Table 2 to com-
pute the monthly cost of natural gas for residential customers. Write a piecewise
definition for the cost of consuming x CCF of natural gas and graph the function.
Table 2 Charges per Month
$0.7675 per CCF for the first 50 CCF
$0.6400 per CCF for the next 150 CCF
$0.6130 per CCF for all over 200 CCF
Exercises 1.2
A In Problems 1–10, find the domain and range of each function. 5. f1x2 = 8 - 1x 6. f1x2 = 5x + 3
2
1. f1x2 = x - 4 2. f1x2 = 1 + 1x 3
7. f1x2 = 27 + 2 x 8. f1x2 = 20 - 18∙ x ∙
2
3. f1x2 = 7 - 2x 4. f1x2 = x + 12 9. f1x2 = 6∙ x ∙ + 9 3
10. f1x2 = - 8 + 2 x
In Problems 11–26, graph each of the functions using the graphs 39. y 40. y
of functions f and g below.
5 5
f(x) g(x)
5 5
x x
25 5 25 5
x x
25 5 25 5
25 25
41. y 42. y
25 25
5
5
11. y = f1x2 + 2 12. y = g1x2 - 1
13. y = f1x + 22 14. y = g1x - 12
x x
15. y = g1x - 32 16. y = f1x + 32 25 5
25 5
17. y = g1x2 - 3 18. y = f1x2 + 3

19. y = - f1x2 20. y = - g1x2 25
25
21. y = 0.5g1x2 22. y = 2f1x2
23. y = 2f1x2 + 1 24. y = - 0.5g1x2 + 3 In Problems 43–48, the graph of the function g is formed by
applying the indicated sequence of transformations to the given
25. y = 21f1x2 + 12 26. y = - 10.5g1x2 + 32
function f. Find an equation for the function g and graph g using
B In Problems 27–34, describe how the graph of each function is - 5 … x … 5 and - 5 … y … 5.
related to the graph of one of the six basic functions in Figure 1 43. The graph of f1x2 = 1x is shifted 3 units to the left and
on page 36. Sketch a graph of each function. 2 units up.
27. g1x2 = - ∙ x + 3 ∙ 28. h1x2 = - ∙ x - 5 ∙ 3
44. The graph of f1x2 = 2 x is shifted 2 units to the right and
29. f1x2 = 1x - 42 2 - 3 30. m1x2 = 1x + 32 2 + 4 3 units down.
31. f1x2 = 7 - 1x 3
32. g1x2 = - 6 + 2 x 45. The graph of f1x2 = ∙ x ∙ is reflected in the x axis and shifted
to the left 3 units.
33. h1x2 = - 3∙ x ∙ 34. m1x2 = - 0.4x2
46. The graph of f1x2 = ∙ x ∙ is reflected in the x axis and shifted
Each graph in Problems 35–42 is the result of applying a se- to the right 1 unit.
quence of transformations to the graph of one of the six basic 47. The graph of f1x2 = x3 is reflected in the x axis and shifted
functions in Figure 1 on page 36. Identify the basic function and 2 units to the right and down 1 unit.
describe the transformation verbally. Write an equation for the
given graph. 48. The graph of f1x2 = x2 is reflected in the x axis and shifted
to the left 2 units and up 4 units.
35. y 36. y
5 5 Graph each function in Problems 49–54.
2 - 2x if x 6 2
49. f1x2 = b
x - 2 if x Ú 2
x x
25 5 25 5
x + 1 if x 6 - 1
50. g1x2 = b
2 + 2x if x Ú - 1
25 25
5 + 0.5x if 0 … x … 10
37. y 38. y 51. h1x2 = b
- 10 + 2x if x 7 10
5 5
10 + 2x if 0 … x … 20
52. h1x2 = b
40 + 0.5x if x 7 20
x x
25 5 25 5
2x if 0 … x … 20
53. h1x2 = c x + 20 if 20 6 x … 40
25 25 0.5x + 40 if x 7 40
Applications
4x + 20 if 0 … x … 20
54. h1x2 = c 2x + 60 if 20 6 x … 100
- x + 360 if x 7 100 67. Price–demand. A retail chain sells bicycle helmets. The
retail price p1x2 (in dollars) and the weekly demand x for a
C Each of the graphs in Problems 55–60 involves a reflection in particular model are related by
the x axis and/or a vertical stretch or shrink of one of the basic
p1x2 = 115 - 41x 9 … x … 289
functions in Figure 1 on page 36. Identify the basic function, and
describe the transformation verbally. Write an equation for the (A) Describe how the graph of function p can be obtained
given graph. from the graph of one of the basic functions in Figure 1
on page 36.
55. y 56. y
5
(B) Sketch a graph of function p using part (A) as an aid.
5
68. Price–supply. The manufacturer of the bicycle helmets in
Problem 67 is willing to supply x helmets at a price of p1x2
x x as given by the equation
25 5 25 5
p1x2 = 41x 9 … x … 289
(A) Describe how the graph of function p can be obtained
25 25 from the graph of one of the basic functions in Figure 1
on page 36.
57. y 58. y
(B) Sketch a graph of function p using part (A) as an aid.
5 5
69. Hospital costs. Using statistical methods, the financial
department of a hospital arrived at the cost equation
x x C1x2 = 0.000481x - 5002 3 + 60,000 100 … x … 1,000
25 5 25 5
where C1x2 is the cost in dollars for handling x cases per month.
(A) Describe how the graph of function C can be obtained
25 25 from the graph of one of the basic functions in Figure 1
on page 36.
59. y 60. y (B) Sketch a graph of function C using part (A) and a
5 5 graphing calculator as aids.
70. Price–demand. A company manufactures and sells in-line
skates. Its financial department has established the price–
x x demand function
25 5 25 5
p1x2 = 190 - 0.0131x - 102 2 10 … x … 100
where p1x2 is the price at which x thousand pairs of in-line
25 25 skates can be sold.
(A) Describe how the graph of function p can be obtained
Changing the order in a sequence of transformations may change from the graph of one of the basic functions in Figure 1
the final result. Investigate each pair of transformations in Problems on page 36.
61–66 to determine if reversing their order can produce a differ-
ent result. Support your conclusions with specific examples and/or (B) Sketch a graph of function p using part (A) and a
mathematical arguments. (The graph of y = f1 - x2 is the reflection graphing calculator as aids.
of y = f1x2 in the y axis.) 71. Electricity rates. Table 3 shows the electricity rates charged
61. Vertical shift; horizontal shift by Origin Energy for the state of Victoria in Australia. The
daily supply charge is a fixed daily charge, independent of
62. Vertical shift; reflection in y axis
the kWh (kilowatt-hours) used during the day.
63. Vertical shift; reflection in x axis
(A) Write a piecewise definition of the daily charge V1x2 for
64. Horizontal shift; vertical stretch a customer who uses x kWh in a day.
65. Horizontal shift; reflection in y axis (B) Graph V1x2.
66. Horizontal shift; horizontal shrink Table 3 Victoria
Daily supply charge $1.34
First 11 kWh at $0.32 per kWh
Over 11 kWh at $0.35 per kWh
72. Electricity rates. Table 4 shows the electricity rates charged (A) Describe how the graph of function w can be obtained
by Origin Energy for the regions in South Australia. from the graph of one of the basic functions in Figure 1,
page 36.
(A) Write a piecewise definition of the daily charge S1x2 for
a customer who uses x kWh in a day. (B) Sketch a graph of function w using part (A) as an aid.
(B) Graph S1x2. 76. Herpetology. The average weight of a particular species of
snake is given by w1x2 = 463x3, 0.2 … x … 0.8, where x is
length in meters and w1x2 is weight in grams.
Table 4 South Australia
(A) Describe how the graph of function w can be obtained
Daily supply charge, $0.90
from the graph of one of the basic functions in Figure 1,
First 11 kWh at $0.40 per kWh
page 36.
Over 11 kWh at $0.43 per kWh
(B) Sketch a graph of function w using part (A) as an aid.
77. Safety research. Under ideal conditions, if a person driving
73. Real estate property tax. Table 5 shows real estate property
a vehicle slams on the brakes and skids to a stop, the speed of
tax rates in Peru.
the vehicle v1x2 (in miles per hour) is given approximately
(A) Write a piecewise definition for T1x2, the amount of tax by v1x2 = C1x, where x is the length of skid marks (in
units due on a real estate’s value of x tax units (1 tax unit feet) and C is a constant that depends on the road condi-
is approximately $1,265). tions and the weight of the vehicle. For a particular vehicle,
v1x2 = 7.081x and 4 … x … 144.
(B) Graph T1x2.
(A) Describe how the graph of function v can be obtained
(C) Find the amount of tax units due on a taxable value of
from the graph of one of the basic functions in Figure 1,
30 tax units. Of 90 tax units.
page 36.
(B) Sketch a graph of function v using part (A) as an aid.
Table 5 Peru Real Estate Property Tax
78. Learning. A production analyst has found that on average
Over But not over Amount of tax units due is
it takes a new person T1x2 minutes to perform a particular
0 tax units 15 tax units 0.2% of the property value
assembly operation after x performances of the operation,
15 tax units 60 tax units 0.03 tax units plus 0.6% of excess 3
where T1x2 = 10 - 2x, 0 … x … 125.
over 15 tax units
60 tax units 0.30 tax units plus 1.0% of (A) Describe how the graph of function T can be obtained
excess over 60 tax units from the graph of one of the basic functions in Figure 1,
page 36.
74. Real estate property tax. Table 6 shows real estate property (B) Sketch a graph of function T using part (A) as an aid.
tax rates in Bolivia.
Answers to Matched Problems
(A) Write a piecewise definition for T1x2, the tax due on a
1. (A) f17292 = 729, h17292 = 531,441,
real estate’s value of x bolivianos ( Bs).
m17292 = 387, 420, 489, n17292 = 27, p17292 = 9,
(B) Graph T1x2.
g17292 = 729
(C) Find the tax due on a taxable value of 300,000 Bs. Of (B) f1 - 5.252 = - 5.25, h1 - 5.252 = 27.5625,
500,000 Bs. Of 1,000,000 Bs.
m1- 5.252 = - 144.7031, n1- 5.252 is not a real number,
p1 - 5.252 = - 1.7380, g1 - 5.252 = 5.25
Table 6 Bolivia Real Estate Property Tax
2. (A) The graph of y = 1x + 5 is the same as the graph
Over But not over Tax due is of y = 1x shifted upward 5 units, and the graph of
Bs 0 Bs 200,000 0.35% of the property value y = 1x - 4 is the same as the graph of y = 1x
Bs 200,000 Bs 400,000 Bs 700 plus 0.50% of excess over shifted downward 4 units. The figure confirms these
Bs 200,000 conclusions.
Bs 400,000 Bs 600,000 Bs 1,700 plus 1.00% of excess
over Bs 400,000 y
Bs 600,000 Bs 3,700 plus 1.50% of excess 10 y 5 Ïx 1 5
over Bs 600,000
5 y 5 Ïx
75. Human weight. A good approximation of the normal x

210 25 5 10
y 5 Ïx 2 4
weight of a person 60 inches or taller but not taller than
25
80 inches is given by w1x2 = 5.5x - 220, where x is height
in inches and w1x2 is weight in pounds. 210
(B) The graph of y = 1x + 5 is the same as the graph y

y5x
of y = 1x shifted to the left 5 units, and the graph 10
of y = 1x - 4 is the same as the graph of y = 1x
shifted to the right 4 units. The figure confirms these 5
conclusions.
x
210 25 5 10
y
25
10 y 5 20.5x
y 5 Ïx 1 5 210
5 y 5 Ïx
y 5 Ïx 2 4 5. The graph of function G is a reflection in the x axis and a
x
210 25 5 10
horizontal translation of 2 units to the left of the graph of
25 y = x3. An equation for G is G1x2 = - 1x + 22 3.
210 6. h(x)
3 3 3 5
3. H1x2 = 2 x + 3, G1x2 = 2 x - 2, M1x2 = 2 x + 2,
3
N1x2 = 2x - 3 4
4. (A) The graph of y = 2x is a vertical stretch of the graph 3
of y = x, and the graph of y = 0.5x is a vertical shrink 2

of the graph of y = x. The figure confirms these
conclusions. 1
x
y 1 2 3 4 5
y 5 2x y5x
10
5
y 5 0.5x 0.7675x if 0 … x … 50
7. C1x2 = µ 38.375 + 0.64 1x - 502 if 50 6 x … 200
x 134.375 + 0.613 1x - 2002 if 200 6 x
210 25 5 10
25
C(x)
210
200
(B) The graph of y = - 0.5x is a vertical shrink and a

reflection in the x axis of the graph of y = x. The figure (200, 134.375)
100
confirms this conclusion.
(50, 38.375)
0 x
100 200 300
1.3 Quadratic Functions

■■ Quadratic Functions, Equations, If the degree of a linear function is increased by one, we obtain a second-degree
and Inequalities function, usually called a quadratic function, another basic function that we will need
■■ Properties of Quadratic Functions in our library of elementary functions. We will investigate relationships between qua-
and Their Graphs dratic functions and the solutions to quadratic equations and inequalities. Other impor-
tant properties of quadratic functions will also be investigated, including maxima and
■■ Applications minima. We will then be in a position to solve important practical problems such as
finding production levels that will generate maximum revenue or maximum profit.
Quadratic Functions, Equations, and Inequalities

The graph of the square function h1x2 = x2 is shown in Figure 1. Notice that the
graph is symmetric with respect to the y axis and that 10, 02 is the lowest point on the
graph. Let’s explore the effect of applying graph transformations to the graph of h.
SECTION 1.3 Quadratic Functions 47
h(x)
5
Indicate how the graph of each function is related to the graph of the function
h1x2 = x2. Find the highest or lowest point, whichever exists, on each graph.
x (A) f1x2 = 1x - 32 2 - 7 = x2 - 6x + 2
25 5
(B) g1x2 = 0.51x + 22 2 + 3 = 0.5x2 + 2x + 5
(C) m1x2 = - 1x - 42 2 + 8 = - x2 + 8x - 8
25 (D) n1x2 = - 31x + 12 2 - 1 = - 3x2 - 6x - 4
Figure 1 Square function h(x) ∙ x 2
The graphs of the functions in Explore and Discuss 1 are similar in shape to the
graph of the square function in Figure 1. All are parabolas. The arc of a basketball
shot is a parabola. Reflecting telescopes, solar furnaces, and automobile headlights
are some of the many applications of parabolas. Each of the functions in Explore and
Discuss 1 is a quadratic function.
DEFINITION Quadratic Functions

If a, b, and c are real numbers with a ∙ 0, then the function
f1x2 = ax2 + bx + c Standard form
is a quadratic function and its graph is a parabola.
CONCEPTUAL INSIGHT
If x is any real number, then ax2 + bx + c is also a real number. According to
the agreement on domain and range in Section 1.1, the domain of a quadratic
function is R, the set of real numbers.
We will discuss methods for determining the range of a quadratic function later
in this section. Typical graphs of quadratic functions are illustrated in Figure 2.
h(x)
Reminder f (x) g(x)
10 10 10
The union of two sets A and B,
denoted AhB, is the set of all ele-
ments that belong to A or B (or both).
So the set of all real numbers x such x x x
25 5
1 - ∞, -24 h 32, ∞ 2.
25 5 25 5
that x2 - 4 Ú 0 (see Fig. 2A) is
210 210 210
(A) f (x) 5 x 2 2 4 (B) g(x) 5 3x 2 2 12x 1 14 (C) h(x) 5 3 2 2x 2 x2
Figure 2 Graphs of quadratic functions
CONCEPTUAL INSIGHT
An x intercept of a function is also called a zero of the function. The x intercept
of a linear function can be found by solving the linear equation y = mx + b = 0
for x, m ∙ 0 (see Section A.2). Similarly, the x intercepts of a quadratic func-
tion can be found by solving the quadratic equation y = ax2 + bx + c = 0
for x, a ∙ 0. Several methods for solving quadratic equations are discussed in

Appendix B, Section B.7. The most popular of these is the quadratic formula.
If ax2 + bx + c = 0, a ∙ 0, then
- b { 2b2 - 4ac
x = , provided b2 - 4ac Ú 0
2a
EXAMPLE 1 Intercepts, Equations, and Inequalities

(A) Sketch a graph of f1x2 = - x2 + 5x + 3 in a rectangular coordinate system.
(B) Find x and y intercepts algebraically to four decimal places.
(C) Graph f1x2 = - x2 + 5x + 3 in a standard viewing window.
(D) Find the x and y intercepts to four decimal places using trace and zero on
your graphing calculator.
(E) Solve the quadratic inequality - x2 + 5x + 3 Ú 0 graphically to four decimal
places using the results of parts (A) and (B) or (C) and (D).
(F) Solve the equation - x2 + 5x + 3 = 4 graphically to four decimal places using
intersect on your graphing calculator.
SOLUTION
(A) Hand-sketch a graph of f by drawing a smooth curve through the plotted points
(Fig. 3).
f (x)
10
x y
-1 -3
0 3 5
1 7
2 9 x
210 25 5 10
3 9
4 7
5 3
6 -3
Figure 3
(B) Find intercepts algebraically:

y intercept: f102 = - 102 2 + 5102 + 3 = 3
x intercepts: f1x2 = - x2 + 5x + 3 = 0 Use the quadratic formula.
- b { 2b2 - 4ac
x = Substitute a = - 1, b = 5, c = 3.
2a
- 152 { 252 - 41 - 12132
x = Simplify.
21 - 12
- 5 { 237
= = - 0.5414 or 5.5414
-2
(C) Use in a graphing calculator (Fig. 4).
10
210 10
210
Figure 4
(D) Find intercepts using a graphing calculator (Fig. 5).
10 10 10
210 10 210 10 210 10
210 210 210
(A) x intercept: 20.5414 (B) x intercept: 5.5414 (C) y intercept: 3

Figure 5
(E) Solve - x2 + 5x + 3 Ú 0 graphically: The quadratic inequality

- x2 + 5x + 3 Ú 0
holds for those values of x for which the graph of f1x2 = - x2 + 5x + 3 in
the figures in parts (A) and (C) is at or above the x axis. This happens for x be-
tween the two x intercepts [found in part (B) or (D)], including the two x inter-
cepts. The solution set for the quadratic inequality is - 0.5414 … x … 5.5414
or 3 - 0.5414, 5.54144.
(F) Solve the equation - x2 + 5x + 3 = 4 using a graphing calculator (Fig. 6).
10 10
210 10 210 10
210 210
(A) 2x 2 1 5x 1 3 5 4 at x 5 0.2087 (B) 2x 2 1 5x 1 3 5 4 at x 5 4.7913

Figure 6
Matched Problem 1
(A) Sketch a graph of g1x2 = 2x2 - 5x - 5 in a rectangular coordinate system.
(B) Find x and y intercepts algebraically to four decimal places.
(C) Graph g1x2 = 2x2 - 5x - 5 in a standard viewing window.
(D) Find the x and y intercepts to four decimal places using trace and the zero
command on your graphing calculator.
(E) Solve 2x2 - 5x - 5 Ú 0 graphically to four decimal places using the results
of parts (A) and (B) or (C) and (D).
(F) Solve the equation 2x2 - 5x - 5 = - 3 graphically to four decimal places
using intersect on your graphing calculator.

How many x intercepts can the graph of a quadratic function have? How many y in-
tercepts? Explain your reasoning.
Properties of Quadratic Functions and Their Graphs

Many useful properties of the quadratic function can be uncovered by transforming
f1x2 = ax2 + bx + c a ∙ 0
into the vertex form
f1x2 = a1x - h2 2 + k
The process of completing the square (see Appendix B.7) is central to the transfor-
mation. We illustrate the process through a specific example and then generalize the
results.
Consider the quadratic function given by
f1x2 = - 2x2 + 16x - 24 (1)
We use completing the square to transform this function into vertex form:
f1x2 = - 2x2 + 16x - 24 Factor the coefficient of x2
out of the first two terms.
= - 21x2 - 8x2 - 24
Add 16 to complete the
square inside the parentheses.
Because of the - 2 outside
= - 21x2 - 8x + ?2 - 24
the parentheses, we have
actually added - 32, so we
must add 32 to the outside.
= - 21x2 - 8x + 162 - 24 + 32 Factor, simplify
The transformation is
2
= - 21x - 42 + 8 complete and can be checked
by multiplying out.
Therefore,
f1x2 = - 21x - 42 2 + 8 (2)
If x = 4, then - 21x - 42 2 = 0 and f142 = 8. For any other value of x, the
negative number - 21x - 42 2 is added to 8, making it smaller. Therefore,
f142 = 8
is the maximum value of f1x2 for all x. Furthermore, if we choose any two x values
that are the same distance from 4, we will obtain the same function value. For ex-
ample, x = 3 and x = 5 are each one unit from x = 4 and their function values are
f132 = - 213 - 42 2 + 8 = 6
f152 = - 215 - 42 2 + 8 = 6
f(x) Therefore, the vertical line x = 4 is a line of symmetry. That is, if the graph of equa-
10
Maximum: f (4) 5 8 tion (1) is drawn on a piece of paper and the paper is folded along the line x = 4,
then the two sides of the parabola will match exactly. All these results are illustrated
by graphing equations (1) and (2) and the line x = 4 simultaneously in the same co-
f(x) 5 22x2 1 16x 2 24
5
5 22(x 2 4)2 1 8 ordinate system (Fig. 7).
From the preceding discussion, we see that as x moves from left to right, f1x2 is
x increasing on (- ∞, 44, and decreasing on 34, ∞ ), and that f1x2 can assume no value
5 10
greater than 8. Thus,
25 Range of f : y … 8 or (- ∞, 84
In general, the graph of a quadratic function is a parabola with line of symmetry
210 parallel to the vertical axis. The lowest or highest point on the parabola, whichever
Line of symmetry: x 5 4 exists, is called the vertex. The maximum or minimum value of a quadratic function
Figure 7 Graph of a quadratic always occurs at the vertex of the parabola. The line of symmetry through the ver-
function tex is called the axis of the parabola. In the example above, x = 4 is the axis of the
parabola and 14, 82 is its vertex.
CONCEPTUAL INSIGHT
Applying the graph transformation properties discussed in Section 1.2 to the
transformed equation,
f1x2 = - 2x2 + 16x - 24
= - 21x - 42 2 + 8
we see that the graph of f1x2 = - 2x2 + 16x - 24 is the graph of g1x2 = x2
vertically stretched by a factor of 2, reflected in the x axis, and shifted to the
right 4 units and up 8 units, as shown in Figure 8.
y
10 g(x) 5 x 2
f(x) 5 22x2 1 16x 2 24

5 5 22(x 2 4)2 1 8
x
25 5 10
25
210
Figure 8 Graph of f is the graph of g transformed
Note the important results we have obtained from the vertex form of the quadratic
function f:
• The vertex of the parabola
• The axis of the parabola
• The maximum value of f1x2
• The range of the function f
• The relationship between the graph of g1x2 = x2 and the graph of
f1x2 = - 2x2 + 16x - 24
The preceding discussion is generalized to all quadratic functions in the following

summary:
SUMMARY Properties of a Quadratic Function and Its Graph

Given a quadratic function and the vertex form obtained by completing the square
f1x2 = ax2 + bx + c a ∙ 0 Standard form
2
= a1x - h2 + k Vertex form
we summarize general properties as follows:
1. The graph of f is a parabola that opens upward if a 7 0, downward if a 6 0
(Fig. 9).
f (x) f (x)
Axis Axis
x5h x5h
k Vertex (h, k)
k Max f (x)
Vertex (h, k) Min f (x)
x x
h h
a.0 a,0
Opens upward Opens downward
Figure 9
2. Vertex: 1h, k2 (parabola increases on one side of the vertex and decreases on
the other)
3. Axis (of symmetry): x = h (parallel to y axis)
4. f1h2 = k is the minimum if a 7 0 and the maximum if a 6 0
5. Domain: All real numbers. Range: (- ∞ , k4 if a 6 0 or 3k, ∞) if a 7 0
6. The graph of f is the graph of g1x2 = ax2 translated horizontally h units and
vertically k units.
EXAMPLE 2 Analyzing a Quadratic Function Given the quadratic function

f1x2 = 0.5 x2 - 6x + 21
(A) Find the vertex form for f.
(B) Find the vertex and the maximum or minimum. State the range of f.
(C) Describe how the graph of function f can be obtained from the graph of
g1x2 = x2 using transformations.
(D) Sketch a graph of function f in a rectangular coordinate system.
(E) Graph function f using a suitable viewing window.
(F) Find the vertex and the maximum or minimum using the appropriate graphing
calculator command.
SOLUTION
(A) Complete the square to find the vertex form:
f1x2 = 0.5 x2 - 6x + 21
= 0.51x2 - 12x + ?2 + 21
= 0.51x2 - 12x + 362 + 21 - 18
= 0.51x - 62 2 + 3
(B) From the vertex form, we see that h = 6 and k = 3. Thus, vertex: 16, 32;
minimum: f162 = 3; range: y Ú 3 or 33, ∞2.
(C) The graph of f1x2 = 0.51x - 62 2 + 3 is the same as the graph of g1x2 = x2
vertically shrunk by a factor of 0.5, and shifted to the right 6 units and up
3 units.
(D) Graph in a rectangular coordinate system (Fig. 10).
f (x)
10
x
5 10
Figure 10
(E) Use a graphing calculator (Fig. 11).
10
210 10
210
Figure 11
(F) Find the vertex and minimum using the minimum command (Fig. 12).
10
210 10
210
Figure 12
Vertex: 16, 32; minimum: f162 = 3
Matched Problem 2 Given the quadratic function f1x2 = - 0.25x2 - 2x + 2

(A) Find the vertex form for f.
(B) Find the vertex and the maximum or minimum. State the range of f.
(C) Describe how the graph of function f can be obtained from the graph of
g1x2 = x2 using transformations.
(D) Sketch a graph of function f in a rectangular coordinate system.
(E) Graph function f using a suitable viewing window.
(F) Find the vertex and the maximum or minimum using the appropriate graphing
calculator command.
Applications
EXAMPLE 3 Maximum Revenue This is a continuation of Example 7 in Section 1.1. Recall
that the financial department in the company that produces a digital camera arrived
at the following price–demand function and the corresponding revenue function:
p1x2 = 94.8 - 5x Price–demand function
R1x2 = xp1x2 = x194.8 - 5x2 Revenue function
where p1x2 is the wholesale price per camera at which x million cameras can be
sold and R(x) is the corresponding revenue (in millions of dollars). Both functions
have domain 1 … x … 15.
(A) Find the value of x to the nearest thousand cameras that will generate the maxi-
mum revenue. What is the maximum revenue to the nearest thousand dollars?
Solve the problem algebraically by completing the square.
(B) What is the wholesale price per camera (to the nearest dollar) that generates the
maximum revenue?
(C) Graph the revenue function using an appropriate viewing window.
(D) Find the value of x to the nearest thousand cameras that will generate the maxi-
mum revenue. What is the maximum revenue to the nearest thousand dollars?
Solve the problem graphically using the maximum command.
SOLUTION
(A) Algebraic solution:
R(x) = x194.8 - 5x2
= - 5x2 + 94.8x
= - 51x2 - 18.96x + ?2
= - 51x2 - 18.96x + 89.87042 + 449.352
= - 51x - 9.482 2 + 449.352
The maximum revenue of 449.352 million dollars 1$449,352,0002 occurs
when x = 9.480 million cameras (9,480,000 cameras).
(B) Finding the wholesale price per camera: Use the price–demand function for an
output of 9.480 million cameras:
p1x2 = 94.8 - 5x
p19.4802 = 94.8 - 519.4802
= $47 per camera
(C) Use a graphing calculator (Fig. 13).
500
0
1 15
Figure 13
(D) Use the maximum command on a graphing calculator (Fig. 14).
500
0
1 15
Figure 14
The manufacture and sale of 9.480 million cameras (9,480,000 cameras) will
generate a maximum revenue of 449.352 million dollars 1$449,352,0002.
Matched Problem 3 The financial department in Example 3, using statistical

and analytical techniques (see Matched Problem 7 in Section 1.1), arrived at the
cost function
C1x2 = 156 + 19.7x Cost function
where C1x2 is the cost (in millions of dollars) for manufacturing and selling x mil-
lion cameras.
(A) Using the revenue function from Example 3 and the preceding cost function,
write an equation for the profit function.
(B) Find the value of x to the nearest thousand cameras that will generate the maxi-
mum profit. What is the maximum profit to the nearest thousand dollars? Solve
the problem algebraically by completing the square.
(C) What is the wholesale price per camera (to the nearest dollar) that generates the
maximum profit?
(D) Graph the profit function using an appropriate viewing window.
(E) Find the output to the nearest thousand cameras that will generate the maxi-
mum profit. What is the maximum profit to the nearest thousand dollars?
Solve the problem graphically using the maximum command.
EXAMPLE 4 Break-Even Analysis Use the revenue function from Example 3 and the cost
function from Matched Problem 3:
R1x2 = x194.8 - 5x2 Revenue function
C1x2 = 156 + 19.7x Cost function
Both have domain 1 … x … 15.
(A) Sketch the graphs of both functions in the same coordinate system.
(B) Break-even points are the production levels at which R1x2 = C1x2. Find the
break-even points algebraically to the nearest thousand cameras.
(C) Plot both functions simultaneously in the same viewing window.
(D) Use intersect to find the break-even points graphically to the nearest thousand
cameras.
(E) Recall that a loss occurs if R1x2 6 C1x2 and a profit occurs if R1x2 7 C1x2.
For what values of x (to the nearest thousand cameras) will a loss occur? A
profit?
SOLUTION
(A) Sketch the functions (Fig. 15).
R(x) C(x)
500 Cost
250 Revenue
x
5 10 15
Figure 15
(B) Find x such that R1x2 = C1x2:

x194.8 - 5x2 = 156 + 19.7x Simplify.
- 5x2 + 75.1x - 156 = 0 Use the quadratic formula.
- 75.1 { 275.12 - 41 - 521 - 1562

x =
21 - 52
- 75.1 { 22,520.01
=
- 10
x = 2.490 and 12.530
The company breaks even at x = 2.490 million cameras (2,490,000 cameras)
and at x = 12.530 million cameras (12,530,000 cameras).
(C) Use a graphing calculator (Fig. 16).
500
Revenue
Cost
0
1 15
Figure 16
(D) Use intersect on a graphing calculator (Fig. 17).
500 500
Revenue Revenue
Cost Cost
0 0
1 15 1 15
(A) (B)
Figure 17
The company breaks even at x = 2.490 million cameras (2,490,000 cameras)
and at x = 12.530 million cameras (12,530,000 cameras).
(E) Use the results from parts (A) and (B) or (C) and (D):
Loss: 1 … x 6 2.490 or 12.530 6 x … 15
Profit: 2.490 6 x 6 12.530
Matched Problem 4 Use the profit equation from Matched Problem 3:

P1x2 = R1x2 - C1x2
= - 5x2 + 75.1x - 156 Profit function
Domain: 1 … x … 15
(A) Sketch a graph of the profit function in a rectangular coordinate system.
(B) Break-even points occur when P1x2 = 0. Find the break-even points algebra-
ically to the nearest thousand cameras.
(C) Plot the profit function in an appropriate viewing window.
(D) Find the break-even points graphically to the nearest thousand cameras.
(E) A loss occurs if P1x2 6 0, and a profit occurs if P1x2 7 0. For what values
of x (to the nearest thousand cameras) will a loss occur? A profit?
A visual inspection of the plot of a data set might indicate that a parabola would be
a better model of the data than a straight line. In that case, rather than using linear regres-
sion to fit a linear model to the data, we would use quadratic regression on a graphing
calculator to find the function of the form y = ax2 + bx + c that best fits the data.
EXAMPLE 5 Outboard Motors Table 1 gives performance data for a boat powered by an
Evinrude outboard motor. Use quadratic regression to find the best model of the
form y = ax2 + bx + c for fuel consumption y (in miles per gallon) as a function
of speed x (in miles per hour). Estimate the fuel consumption (to one decimal place)
at a speed of 12 miles per hour.
Table 1
rpm mph mpg
2,500 10.3 4.1
3,000 18.3 5.6
3,500 24.6 6.6
4,000 29.1 6.4
4,500 33.0 6.1
5,000 36.0 5.4
5,400 38.9 4.9
SOLUTION Enter the data in a graphing calculator (Fig. 18A) and find the quadratic
regression equation (Fig. 18B). The data set and the regression equation are graphed
in Figure 18C. Using trace, we see that the estimated fuel consumption at a speed of
12 mph is 4.5 mpg.
10
0
0 50
(A) (B) (C)

Figure 18
Matched Problem 5 Refer to Table 1. Use quadratic regression to find the

best model of the form y = ax2 + bx + c for boat speed y (in miles per hour) as a
function of engine speed x (in revolutions per minute). Estimate the boat speed (in
miles per hour, to one decimal place) at an engine speed of 3,400 rpm.
Exercises 1.3
A In Problems 1–8, find the vertex form of each quadratic function 15. Function n in the figure for Problem 13
by completing the square.
16. Function m in the figure for Problem 14
1. f1x2 = x2 - 10x 2. f1x2 = x2 + 16x
17. Function f in the figure for Problem 13
3. f1x2 = x2 + 20x + 50 4. f1x2 = x2 - 12x - 8
18. Function g in the figure for Problem 14
5. f1x2 = - 2x2 + 4x - 5 6. f1x2 = 4x2 - 24x + 26
In Problems 19–22, find each of the following:
7. f1x2 = 2x2 + 2x + 1 8. f1x2 = - 5x2 + 15x - 11
(A) Intercepts (B) Vertex
In Problems 9–12, write a brief verbal description of the relation-
(C) Maximum or minimum (D) Range
ship between the graph of the indicated function and the graph of
y = x 2. 19. f1x2 = - 1x - 32 2 + 2 20. g1x2 = - 1x + 22 2 + 3
2 2
9. f1x2 = x - 4x + 3 10. g1x2 = x - 2x - 5 21. m1x2 = 1x + 12 2 - 2 22. n1x2 = 1x - 42 2 - 3
2 2
11. m1x2 = - x + 10x - 16 12. n1x2 = - x + 8x - 9
B In Problems 23–26, write an equation for each graph in the form
13. Match each equation with a graph of one of the functions f, g, m, y = a1x - h2 2 + k, where a is either 1 or - 1 and h and k are
or n in the figure. integers.
(A) y = - 1x + 22 2 + 1 (B) y = 1x - 22 2 - 1 23. 24.
y y
(C) y = 1x + 22 2 - 1 (D) y = - 1x - 22 2 + 1
5 5
y
f g
x x
25 5 5
x
25 5
25 25
25. y 26. y
m n
5 5
14. Match each equation with a graph of one of the functions
f, g, m, or n in the figure.
(A) y = 1x - 32 2 - 4 (B) y = - 1x + 32 2 + 4 x
5 x
25 5
(C) y = - 1x - 32 2 + 4 (D) y = 1x + 32 2 - 4
y
f g 25
5 25
In Problems 27–32, find the vertex form for each quadratic func-
tion. Then find each of the following:
x
27 7
25
m n 27. f1x2 = x2 - 8x + 12 28. g1x2 = x2 - 12x + 27
For the functions indicated in Problems 15–18, find each of the 29. r1x2 = - 4x2 + 16x - 15 30. s1x2 = - 4x2 - 8x - 3
following to the nearest integer by referring to the graphs for 31. u1x2 = 0.5x2 - 2x + 5 32. v1x2 = 0.5x2 + 6x + 19
Problems 13 and 14.
33. Let f1x2 = 0.3x2 - x - 8. Solve each equation graphically
(A) Intercepts (B) Vertex to two decimal places.
(C) Maximum or minimum (D) Range (A) f1x2 = 4 (B) f1x2 = - 1 (C) f1x2 = - 9
34. Let g1x2 = - 0.6x2 + 3x + 4. Solve each equation graphi- 56. f1x2 = - 0.7x2 + 6.3x
cally to two decimal places.
g1x2 = 1.1x + 4.8
(A) g1x2 = - 2 (B) g1x2 = 5 (C) g1x2 = 8
0 … x … 9
35. Let f1x2 = 125x - 6x2. Find the maximum value of f to
four decimal places graphically. 57. How can you tell from the graph of a quadratic function
2 whether it has exactly one real zero?
36. Let f1x2 = 100x - 7x - 10. Find the maximum value of f
to four decimal places graphically. 58. How can you tell from the graph of a quadratic function
whether it has no real zeros?
C In Problems 37–40, first write each function in vertex form; then
find each of the following (to two decimal places): 59. How can you tell from the standard form y = ax2 + bx + c
whether a quadratic function has two real zeros?
60. How can you tell from the standard form y = ax2 + bx + c
(C) Maximum or minimum (D) Range whether a quadratic function has exactly one real zero?
37. g1x2 = 0.25x2 - 1.5x - 7 61. How can you tell from the vertex form y = a1x - h2 2 + k
whether a quadratic function has no real zeros?
38. m1x2 = 0.20x2 - 1.6x - 1
62. How can you tell from the vertex form y = a1x - h2 2 + k
39. f1x2 = - 0.12x2 + 0.96x + 1.2
whether a quadratic function has two real zeros?
40. n1x2 = - 0.25x2 - 4.50x - 9.75
In Problems 63 and 64, assume that a, b, c, h, and k are constants
In Problems 41–44, use interval notation to write the solution set with a ∙ 0 such that
of the inequality.
ax2 + bx + c = a1x - h2 2 + k
41. 1x - 32 1x + 52 7 0 42. 1x + 62 1x - 32 6 0 for all real numbers x.
2 2
43. x + x - 6 … 0 44. x + 9x + 14 Ú 0
b 4ac - b2
63. Show that h = - . 64. Show that k = .
Solve Problems 45–50 graphically to two decimal places using a 2a 4a
graphing calculator.
45. 2 - 5x - x2 = 0 46. 7 + 3x - 2x2 = 0
47. 1.9x2 - 1.5x - 5.6 6 0 48. 3.4 + 2.9x - 1.1x2 Ú 0
Applications
49. 2.8 + 3.1x - 0.9x2 … 0 50. 1.8x2 - 3.1x - 4.9 7 0 65. Tire mileage. An automobile tire manufacturer collected the
data in the table relating tire pressure x (in pounds per square
51. Given that f is a quadratic function with maximum
inch) and mileage (in thousands of miles):
f1x2 = f1 - 42 = - 9, find the axis, vertex, range, and x
intercepts. x Mileage
52. Given that f is a quadratic function with minimum 28 45
f1x2 = f132 = 7, find the axis, vertex, range, and x 30 52
intercepts. 32 55
34 51
In Problems 53–56, 36 47
(A) Graph f and g in the same coordinate system.
A mathematical model for the data is given by
(B) Solve f1x2 = g1x2 algebraically to two decimal places.
f1x2 = - 0.518x2 + 33.3x - 481
(C) Solve f1x2 7 g1x2 using parts (A) and (B).
(A) Complete the following table. Round values of f(x) to
(D) Solve f1x2 6 g1x2 using parts (A) and (B). one decimal place.
53. f1x2 = - 0.4x1x - 102 x Mileage f(x)

28 45
g1x2 = 0.3x + 5
30 52
0 … x … 10 32 55
34 51
54. f1x2 = - 0.7x1x - 72
36 47
g1x2 = 0.5x + 3.5
(B) Sketch the graph of f and the mileage data in the same
0 … x … 7 coordinate system.
55. f1x2 = - 0.9x2 + 7.2x (C) Use the modeling function f(x) to estimate the mileage
g1x2 = 1.2x + 5.5 for a tire pressure of 31 lbs/sq in. and for 35 lbs/sq in.
Round answers to two decimal places.
0 … x … 8
(D) Write a brief description of the relationship between tire (A) Sketch a graph of the revenue function in a rectangular
pressure and mileage. coordinate system.
66. Automobile production. The table shows the retail market (B) Find the value of x that will produce the maximum rev-
share of passenger cars (excluding minivehicles) from Toyota enue. What is the maximum revenue?
Motor Corporation as a percentage of the Japanese market.
(C) What is the wholesale price per chip that produces the
Year Market Share maximum revenue?
1985 41.8% 70. Revenue. The marketing research department for a company
1990 41.9% that manufactures and sells notebook computers established
1995 40.0% the following price–demand and revenue functions:
2000 43.2%
p1x2 = 2,000 - 60x Price–demand function
2005 43.6%
2010 48.5% R1x2 = xp1x2 Revenue function
A mathematical model for this data is given by = x12,000 - 60x2
f1x2 = 0.0237x2 - 0.59x + 44.52 where p(x) is the wholesale price in dollars at which x
where x = 0 corresponds to 1980. thousand computers can be sold, and R(x) is in thousands of
dollars. Both functions have domain 1 … x … 25.
(A) Complete the following table. Round values of f(x) to
two decimal places. (A) Sketch a graph of the revenue function in a rectangular
coordinate system.
x Market Share f(x)
(B) Find the value of x that will produce the maximum
5 41.8
revenue. What is the maximum revenue to the nearest
10 41.9
thousand dollars?
15 40.0
20 43.2 (C) What is the wholesale price per computer (to the nearest
25 43.6 dollar) that produces the maximum revenue?
30 48.5 71. Break-even analysis. Use the revenue function from Prob-
lem 69 and the given cost function:
(B) Sketch the graph of f and the market share data in the
same coordinate system. R1x2 = x175 - 3x2 Revenue function
(C) Use values of the modeling function f to estimate C1x2 = 125 + 16x Cost function
Toyota’s market share in 2020 and in 2023.
where x is in millions of chips, and R(x) and C(x) are in mil-
(D) Write a brief verbal description of Toyota’s market share lions of dollars. Both functions have domain 1 … x … 20.
from 1985 to 2010. (A) Sketch a graph of both functions in the same rectangular
67. Tire mileage. Using quadratic regression on a graphing coordinate system.
calculator, show that the quadratic function that best fits the (B) Find the break-even points to the nearest thousand chips.
data on tire mileage in Problem 65 is
(C) For what values of x will a loss occur? A profit?
f1x2 = - 0.518x2 + 33.3x - 481
72. Break-even analysis. Use the revenue function from
68. Automobile production. Using quadratic regression on a Problem 70, and the given cost function:
graphing calculator, show that the quadratic function that best
fits the data on market share in Problem 66 is R1x2 = x12,000 - 60x2 Revenue function
f1x2 = 0.0237x2 - 0.59x + 44.52 C1x2 = 4,000 + 500x Cost function
69. Revenue. The marketing research department for a company where x is thousands of computers, and C(x) and R(x)
that manufactures and sells memory chips for microcomputers are in thousands of dollars. Both functions have domain
established the following price–demand and revenue 1 … x … 25.
functions: (A) Sketch a graph of both functions in the same rectangular
coordinate system.
p1x2 = 75 - 3x Price–demand function
R1x2 = xp1x2 = x175 - 3x2 Revenue function (B) Find the break-even points.
where p(x) is the wholesale price in dollars at which x million (C) For what values of x will a loss occur? A profit?
chips can be sold, and R(x) is in millions of dollars. Both 73. Profit-loss analysis. Use the revenue and cost functions
functions have domain 1 … x … 20. from Problem 71:
R1x2 = x175 - 3x2 Revenue function
C1x2 = 125 + 16x Cost function
where x is in millions of chips, and R(x) and C(x) are in mil- Table for 77 and 78
lions of dollars. Both functions have domain 1 … x … 20. rpm mph mpg
(A) Form a profit function P, and graph R, C, and P in the 1,500 4.5 8.2
same rectangular coordinate system.
2,000 5.7 6.9
(B) Discuss the relationship between the intersection points 2,500 7.8 4.8
of the graphs of R and C and the x intercepts of P. 3,000 9.6 4.1
(C) Find the x intercepts of P and the break-even points to 3,500 13.4 3.7
the nearest thousand chips.
(D) Find the value of x (to the nearest thousand chips) that 78. Outboard motors. The table gives performance data for a
produces the maximum profit. Find the maximum profit boat powered by an Evinrude outboard motor. Find a quadratic
(to the nearest thousand dollars), and compare with regression model 1y = ax2 + bx + c2 for fuel consumption
Problem 69B. y (in miles per gallon) as a function of engine speed (in revolu-
tions per minute). Estimate the fuel consumption at an engine
74. Profit-loss analysis. Use the revenue function from Problem speed of 2,300 revolutions per minute.
70 and the given cost function:
R1x2 = x12,000 - 60x2 Revenue function Answers to Matched Problems
C1x2 = 4,000 + 500x Cost function 1. (A) g(x)
where x is thousands of computers, and R(x) and C(x) 10

are in thousands of dollars. Both functions have domain
1 … x … 25.
x
(A) Form a profit function P, and graph R, C, and P in the 210 10
same rectangular coordinate system.
210
(B) Discuss the relationship between the intersection points
of the graphs of R and C and the x intercepts of P. (B) x intercepts: - 0.7656, 3.2656; y intercept: - 5
(C) Find the x intercepts of P and the break-even points. (C) 10
(D) Find the value of x that produces the maximum profit.

Find the maximum profit and compare with Problem 70B. 210 10
75. Medicine. The French physician Poiseuille was the first to
discover that blood flows faster near the center of an artery
210
than near the edge. Experimental evidence has shown that the
rate of flow v (in centimeters per second) at a point x centime- (D) x intercepts: - 0.7656, 3.2656; y intercept: - 5
ters from the center of an artery (see the figure) is given by
(E) x … - 0.7656 or x Ú 3.2656; or 1 - ∞, - 0.76564 or
v = f1x2 = 1,00010.04 - x2 2 0 … x … 0.2 33.2656, ∞2
Find the distance from the center that the rate of flow is 20 centi- (F) x = - 0.3508, 2.8508
meters per second. Round answer to two decimal places. 2. (A) f1x2 = - 0.251x + 42 2 + 6.
(B) Vertex: 1 - 4, 62; maximum: f1 - 42 = 6; range: y … 6
Artery
or 1 -∞ , 64
(C) The graph of f1x2 = - 0.251x + 42 2 + 6 is the same
x y as the graph of g1x2 = x2 vertically shrunk by a factor
of 0.25, reflected in the x axis, and shifted 4 units to the
left and 6 units up.
(D) y
Figure for 75 and 76
10
76. Medicine. Refer to Problem 75. Find the distance from the
center that the rate of flow is 30 centimeters per second.
x
Round answer to two decimal places. 210 10
77. Outboard motors. The table gives performance data for a

210
boat powered by an Evinrude outboard motor. Find a qua-
dratic regression model 1y = ax2 + bx + c2 for boat speed
y (in miles per hour) as a function of engine speed (in revolu-
tions per minute). Estimate the boat speed at an engine speed
of 3,100 revolutions per minute.
(E) 10 The manufacture and sale of 7,510,000 cameras will produce

a maximum profit of $126,001,000. (Notice that maximum
profit does not occur at the same value of x where maximum
210 10 revenue occurs.)
4. (A) P(x)
210
300
x
(F) Vertex: 1 - 4, 62; maximum: f1 - 42 = 6 0 5 10 15
10
2300
210 10 (B) x = 2.490 million cameras (2,490,000 cameras) and

x = 12.530 million cameras (12,530,000 cameras)
(C) 300
210
3. (A) P1x2 = R1x2 - C1x2 = - 5x2 + 75.1x - 156

1 15
(B) P1x2 = R1x2 - C1x2 = - 51x - 7.512 2 + 126.0005;
the manufacture and sale of 7,510,000 cameras will
produce a maximum profit of $126,001,000. 2300
(C) p17.5102 = $57
(D) 300 (D) x = 2.490 million cameras (2,490,000 cameras) and
x = 12.530 million cameras (12,530,000 cameras)
1 15
(E) Loss: 1 … x 6 2.490 or 12.530 6 x … 15; profit:
2.490 6 x 6 12.530
5.
2300
(E) 300
1 15
22.9 mph
2300
1.4 Polynomial and Rational Functions

■■ Polynomial Functions Linear and quadratic functions are special cases of the more general class of
■■ Regression Polynomials
polynomial functions. Polynomial functions are a special case of an even larger class
of functions, the rational functions. We will describe the basic features of the graphs
■■ Rational Functions of polynomial and rational functions. We will use these functions to solve real-world
■■ Applications problems where linear or quadratic models are inadequate, for example, to determine
the relationship between length and weight of a species of fish, or to model the train-
ing of new employees.
Polynomial Functions
A linear function has the form f1x2 = mx + b 1where m ∙ 02 and is a polyno-
mial function of degree 1. A quadratic function has the form f1x2 = ax2 + bx + c
1where a ∙ 02 and is a polynomial function of degree 2. Here is the general defini-
tion of a polynomial function.
SECTION 1.4 Polynomial and Rational Functions 63
DEFINITION Polynomial Function

A polynomial function is a function that can be written in the form
f1x2 = anx n + an - 1x n - 1 + g + a1x + a0
for n a nonnegative integer, called the degree of the polynomial. The coefficients
a0, a1, . . . , an are real numbers with an ∙ 0. The domain of a polynomial function
is the set of all real numbers.
Figure 1 shows graphs of representative polynomial functions of degrees 1 through 6.

The figure, which also appears on the inside back cover, suggests some general prop-
erties of graphs of polynomial functions.
f (x) g(x) h(x)
5 5 5
x x x
25 5 25 5 25 5
25 25 25
(A) f (x) 5 x 2 2 (B) g(x) 5 x3 2 2x (C) h(x) 5 x5 2 5x3 1 4x 1 1
F(x) G(x) H(x)
5 5 5
x x x
25 5 25 5 25 5
25 25 25
(D) F(x) 5 x2 2 2x 1 2 (E) G(x) 5 2x 4 2 4x 2 1 x 2 1 (F) H(x) 5 x6 2 7x4 1 14x2 2 x 2 5
Figure 1 Graphs of polynomial functions
Notice that the odd-degree polynomial graphs start negative, end positive, and
cross the x axis at least once. The even-degree polynomial graphs start positive, end
positive, and may not cross the x axis at all. In all cases in Figure 1, the leading coef-
ficient—that is, the coefficient of the highest-degree term—was chosen positive. If
any leading coefficient had been chosen negative, then we would have a similar graph
but reflected in the x axis.
A polynomial of degree n can have, at most, n linear factors. Therefore, the graph
Reminder of a polynomial function of positive degree n can intersect the x axis at most n times.
Note from Figure 1 that a polynomial of degree n may intersect the x axis fewer than
Only real numbers can be x intercepts.
n times. An x intercept of a function is also called a zero or root of the function.
Functions may have complex zeros
The graph of a polynomial function is continuous, with no holes or breaks. That
that are not real numbers, but such
zeros, which are not x intercepts, will is, the graph can be drawn without removing a pen from the paper. Also, the graph
not be discussed in this book. of a polynomial has no sharp corners. Figure 2 shows the graphs of two functions—
one that is not continuous, and the other that is continuous but with a sharp corner.
Neither function is a polynomial.
f (x) h(x)
x x
Discontinuous–break
at x 5 0
Continuous, but sharp
corner at (0, 23)
2|x|
(A) f (x) 5 (B) h(x) 5 | x | 2 3
x
Figure 2 Discontinuous and sharp-corner functions
Regression Polynomials
In Section A.3, we saw that regression techniques can be used to fit a straight line to a
set of data. Linear functions are not the only ones that can be applied in this manner.
Most graphing calculators have the ability to fit a variety of curves to a given set of
data. We will discuss polynomial regression models in this section and other types of
regression models in later sections.
EXAMPLE 1 Estimating the Weight of a Fish Using the length of a fish to estimate its weight
is of interest to both scientists and sport anglers. The data in Table 1 give the average
weights of lake trout for certain lengths. Use the data and regression techniques to
find a polynomial model that can be used to estimate the weight of a lake trout for
any length. Estimate (to the nearest ounce) the weights of lake trout of lengths 39,
40, 41, 42, and 43 inches, respectively.
Table 1 Lake Trout
Length (in.) Weight (oz) Length (in.) Weight (oz)
x y x y
10 5 30 152
14 12 34 226
18 26 38 326
22 56 44 536
26 96
SOLUTION The graph of the data in Table 1 (Fig. 3A) indicates that a linear regres-
sion model would not be appropriate in this case. And, in fact, we would not expect
a linear relationship between length and weight. Instead, it is more likely that the
weight would be related to the cube of the length. We use a cubic regression polyno-
mial to model the data (Fig. 3B). Figure 3C adds the graph of the polynomial model
to the graph of the data. The graph in Figure 3C shows that this cubic polynomial
600
0 50
0
(A) (B)
Figure 3
600
0 50
0
(C) (D)
does provide a good fit for the data. (We will have more to say about the choice of
functions and the accuracy of the fit provided by regression analysis later in the
book.) Figure 3D shows the estimated weights for the lengths requested.
Matched Problem 1 The data in Table 2 give the average weights of pike for
certain lengths. Use a cubic regression polynomial to model the data. Estimate (to
the nearest ounce) the weights of pike of lengths 39, 40, 41, 42, and 43 inches,
respectively.
Table 2 Pike
Length (in.) Weight (oz) Length (in.) Weight (oz)
x y x y
10 5 30 108
14 12 34 154
18 26 38 210
22 44 44 326
26 72 52 522
Rational Functions
Just as rational numbers are defined in terms of quotients of integers, rational functions
are defined in terms of quotients of polynomials. The following equations specify ratio-
nal functions:
1 x - 2 x3 - 8
f1x2 = g1x2 = h1x2 =
x 2
x - x - 6 x
p1x2 = 3x2 - 5x q1x2 = 7 r1x2 = 0
DEFINITION Rational Function

A rational function is any function that can be written in the form
n1x2
f1x2 = d1x2 ∙ 0
d1x2
where n1x2 and d1x2 are polynomials. The domain is the set of all real numbers
such that d1x2 ∙ 0.
Figure 4 shows the graphs of representative rational functions. Note, for ex-
ample, that in Figure 4A the line x = 2 is a vertical asymptote for the function. The
graph of f gets closer to this line as x gets closer to 2. The line y = 1 in Figure 4A
is a horizontal asymptote for the function. The graph of f gets closer to this line as x
increases or decreases without bound.
f (x) f (x) f (x)

5 5 5
x x x
25 5 25 5 25 5
25 25 25
x23 8 1
(A) f (x) 5 (B) f (x) 5 2 (C) f (x) 5 x 1
x22 x 24 x
Figure 4 Graphs of rational functions
The number of vertical asymptotes of a rational function f1x2 = n1x2 >d1x2 is

at most equal to the degree of d1x2. A rational function has at most one horizontal
asymptote (note that the graph in Fig. 4C does not have a horizontal asymptote).
Moreover, the graph of a rational function approaches the horizontal asymptote
(when one exists) both as x increases and decreases without bound.
EXAMPLE 2 Graphing Rational Functions Given the rational function
3x
f1x2 = 2
x - 4
(A) Find the domain.
(B) Find the x and y intercepts.
(C) Find the equations of all vertical asymptotes.
(D) If there is a horizontal asymptote, find its equation.
(E) Using the information from (A)–(D) and additional points as necessary, sketch
a graph of f.
SOLUTION
(A) x2 - 4 = 1x - 221x + 22, so the denominator is 0 if x = - 2 or x = 2.
Therefore the domain is the set of all real numbers except - 2 and 2.
(B) x intercepts: f1x2 = 0 only if 3x = 0, or x = 0. So the only x intercept is 0.
y intercept:
3#0 0
f102 = 2 = = 0
0 - 4 -4
So the y intercept is 0.
(C) Consider individually the values of x for which the denominator is 0, namely,
2 and - 2, found in part (A).
(i) If x = 2, the numerator is 6, and the denominator is 0, so f122 is undefined.
But for numbers just to the right of 2 1like 2.1, 2.01, 2.0012, the numera-
tor is close to 6, and the denominator is a positive number close to 0, so
the fraction f1x2 is large and positive. For numbers just to the left of 2
1like 1.9, 1.99, 1.9992, the numerator is close to 6, and the denominator is
a negative number close to 0, so the fraction f1x2 is large (in absolute value)
and negative. Therefore, the line x = 2 is a vertical asymptote, and f1x2 is
positive to the right of the asymptote, and negative to the left.
(ii) If x = - 2, the numerator is - 6, and the denominator is 0, so f122 is unde-
fined. But for numbers just to the right of - 2 1like - 1.9, - 1.99, - 1.9992,
the numerator is close to - 6, and the denominator is a negative number
close to 0, so the fraction f1x2 is large and positive. For numbers just to
the left of - 2 1like - 2.1, - 2.01, - 2.0012, the numerator is close to - 6,
and the denominator is a positive number close to 0, so the fraction f1x2
is large (in absolute value) and negative. Therefore, the line x = - 2 is a
vertical asymptote, and f1x2 is positive to the right of the asymptote and
negative to the left.
(D) Rewrite f1x2 by dividing each term in the numerator and denominator by the
highest power of x in f1x2.
3x 3
3x x2 x
f1x2 = 2= 2 =
x - 4 x 4 4
2 - 2
1 - 2
x x x
As x increases or decreases without bound, the numerator tends to 0 and the
denominator tends to 1; so, f1x2 tends to 0. The line y = 0 is a horizontal
asymptote.
(E) Use the information from parts (A)–(D) and plot additional points as necessary
to complete the graph, as shown in Figure 5.
f (x)
x f 1 x2
-4 -1
5 -2.3 -5.3
-1.7 4.6
25
x 0 0
5
1.7 -4.6
25 2.3 5.3
4 1
Figure 5
3x + 3
Matched Problem 2 Given the rational function g1x2 =
x2 - 9
(A) Find the domain.
(B) Find the x and y intercepts.
(C) Find the equations of all vertical asymptotes.
(D) If there is a horizontal asymptote, find its equation.
(E) Using the information from parts (A)–(D) and additional points as necessary,
sketch a graph of g.
CONCEPTUAL INSIGHT
Consider the rational function
3x2 - 12x 3x1x - 42
g1x2 = 3 2 = 2
x - 4x - 4x + 16 1x - 421x - 42
The numerator and denominator of g have a common zero, x = 4. If x ∙ 4, then
we can cancel the factor x - 4 from the numerator and denominator, leaving the
function f1x2 of Example 2. So the graph of g (Fig. 6) is identical to the graph
of f (Fig. 5), except that the graph of g has an open dot at 14, 12, indicating that
4 is not in the domain of g. In particular, f and g have the same asymptotes. Note
that the line x = 4 is not a vertical asymptote of g, even though 4 is a zero of its
denominator.
g(x)
25
x
5
25
Figure 6
Graphing rational functions is aided by locating vertical and horizontal asymp-

totes first, if they exist. The following general procedure is suggested by Example 2
and the Conceptual Insight above.
PROCEDURE Vertical and Horizontal Asymptotes of Rational Functions

Consider the rational function
n1x2
f1x2 =
d1x2
where n1x2 and d1x2 are polynomials.
Vertical asymptotes:
Case 1. Suppose n1x2 and d1x2 have no real zero in common. If c is a real number
such that d1c2 = 0, then the line x = c is a vertical asymptote of the graph of f.
Case 2. If n1x2 and d1x2 have one or more real zeros in common, cancel common
linear factors, and apply Case 1 to the reduced function. (The reduced function has
the same asymptotes as f.)
Horizontal asymptote:
Case 1. If degree n1x2 6 degree d1x2, then y = 0 is the horizontal asymptote.
Case 2. If degree n1x2 = degree d1x2, then y = a>b is the horizontal asymptote,
where a is the leading coefficient of n1x2, and b is the leading coefficient of d1x2.
Case 3. If degree n1x2 7 degree d1x2, there is no horizontal asymptote.
Example 2 illustrates Case 1 of the procedure for horizontal asymptotes. Cases 2

and 3 are illustrated in Example 3 and Matched Problem 3.
EXAMPLE 3 Finding Asymptotes Find the vertical and horizontal asymptotes of the rational
function
3x2 + 3x - 6
f1x2 =
2x2 - 2
SOLUTION Vertical asymptotes We factor the numerator n1x2 and the denominator
d1x2:
n1x2 = 3 1 x2 + x - 2 2 = 31x - 121x + 22
d1x2 = 2 1 x2 - 1 2 = 21x - 121x + 12
The reduced function is
31x + 22
21x + 12
which, by the procedure, has the vertical asymptote x = - 1. Therefore, x = - 1 is
the only vertical asymptote of f.
Horizontal asymptote Both n1x2 and d1x2 have degree 2 (Case 2 of the procedure
for horizontal asymptotes). The leading coefficient of the numerator n1x2 is 3, and
the leading coefficient of the denominator d1x2 is 2. So y = 3>2 is the horizontal
asymptote.
Matched Problem 3 Find the vertical and horizontal asymptotes of the rational
function
x3 - 4x
f1x2 =
x2 + 5x

A function f is bounded if the entire graph of f lies between two horizontal lines. The
only polynomials that are bounded are the constant functions, but there are many
rational functions that are bounded. Give an example of a bounded rational function,
with domain the set of all real numbers, that is not a constant function.
Applications
Rational functions occur naturally in many types of applications.
EXAMPLE 4 Employee Training A company that manufactures computers has established that,
on the average, a new employee can assemble N1t2 components per day after t days
of on-the-job training, as given by
50t
N1t2 = t Ú 0
t + 4
Sketch a graph of N, 0 … t … 100, including any vertical or horizontal asymp-
totes. What does N1t2 approach as t increases without bound?
N(t)
SOLUTION Vertical asymptotes None for t Ú 0
Horizontal asymptote
50
50t 50
N1t2 = =
t + 4 4
1 +
t
N1t2 approaches 50 (the leading coefficient of 50t divided by the leading coeffi-
t
cient of t + 4) as t increases without bound. So y = 50 is a horizontal asymptote.
0 50 100
Sketch of graph Note that N(0) = 0, N(25) ≈ 43, and N(100) ≈ 48. We draw a
Figure 7 smooth curve through (0, 0), (25, 43) and (100, 48) (Fig. 7).
N1t2 approaches 50 as t increases without bound. It appears that 50 components
per day would be the upper limit that an employee would be expected to assemble.
25t + 5
Matched Problem 4 Repeat Example 4 for N1t2 = t Ú 0.
t + 5
Exercises 1.4
A In Problems 1–10, for each polynomial function find the following: 17. f (x) 18. f (x)
(A) Degree of the polynomial 5 5
(B) All x intercepts

(C) The y intercept
x x
25 5 25 5
2
1. f1x2 = 7x + 21 2. f1x2 = x - 5x + 6
2
3. f1x2 = x + 9x + 20 4. f1x2 = 30 - 3x
25 25
2 6 4
5. f1x2 = x + 2x + 3x + 15
6. f1x2 = 5x4 + 3x2 + x10 + 7x8 + 9 19. What is the maximum number of x intercepts that a polyno-
mial of degree 10 can have?
7. f1x2 = x2 1x + 62 3
20. What is the maximum number of x intercepts that a polyno-
8. f1x2 = 1x - 82 2 1x + 42 3 mial of degree 11 can have?
9. f1x2 = 1x2 - 252 1x3 + 82 3 21. What is the minimum number of x intercepts that a polyno-
2 2 4 mial of degree 9 can have? Explain.
10. f1x2 = 12x - 52 1x - 92
Each graph in Problems 11–18 is the graph of a polynomial func- 22. What is the minimum number of x intercepts that a polyno-
tion. Answer the following questions for each graph: mial of degree 6 can have? Explain.
(A) What is the minimum degree of a polynomial function that B For each rational function in Problems 23–28,
could have the graph? (A) Find the intercepts for the graph.
(B) Is the leading coefficient of the polynomial negative or (B) Determine the domain.
positive?
(C) Find any vertical or horizontal asymptotes for the graph.
11. f(x) 12. f (x)
(D) Sketch any asymptotes as dashed lines. Then sketch a graph
5 5 of y = f1x2.
x + 2 x - 3
23. f1x2 = 24. f1x2 =
x x x - 2 x + 3
25 5 25 5
3x 2x
25. f1x2 = 26. f1x2 =
x + 2 x - 3
25 25 4 - 2x 3 - 3x
27. f1x2 = 28. f1x2 =
x - 4 x - 2
13. f(x) 14. f (x)
29. Compare the graph of y = 2x4 to the graph of
5 5 y = 2x4 - 5x2 + x + 2 in the following two viewing
windows:
(A) - 5 … x … 5, - 5 … y … 5
x x
25 5 25 5 (B) - 5 … x … 5, - 500 … y … 500
30. Compare the graph of y = x3 to the graph of
y = x3 - 2x + 2 in the following two viewing windows:
25 25 (A) - 5 … x … 5, - 5 … y … 5
(B) - 5 … x … 5, - 500 … y … 500
15. f(x) 16. f (x)
31. Compare the graph of y = - x5 to the graph of
5 5 y = - x5 + 4x3 - 4x + 1 in the following two viewing
windows:
(A) - 5 … x … 5, - 5 … y … 5
x x
25 5 25 5 (B) - 5 … x … 5, - 500 … y … 500
25 25
32. Compare the graph of y = - x5 to the graph of 54. Write an equation for the lowest-degree polynomial function
y = - x5 + 5x3 - 5x + 2 in the following two viewing with the graph and intercepts shown in the figure.
windows: f (x)
(A) - 5 … x … 5, - 5 … y … 5
5
(B) - 5 … x … 5, - 500 … y … 500
In Problems 33–40, find the equation of any horizontal asymptote.
x
25 5
4x3 + x2 - 10 6x4 - x3 + 2
33. f1x2 = 3 34. f1x2 = 4
5x + 7x2 + 9 4x + 10x + 5
1 - 5x + x2 8 - x3 25
35. f1x2 = 36. f1x2 =
2 + 3x + 4x2 1 + 2x3
55. Write an equation for the lowest-degree polynomial function
x4 + 2x2 + 1 3 + 5x with the graph and intercepts shown in the figure.
37. f1x2 = 38. f1x2 = 2
1 - x5 x + x + 3
f (x)
x2 + 6x + 1 x2 + x4 + 1
39. f1x2 = 40. f1x2 = 3 5
x - 5 x + 2x - 4
In Problems 41–46, find the equations of any vertical asymptotes.
x
x2 + 1 x2 - 16 25 5
41. f1x2 = 42. f1x2 =
1x2 - 121x2 - 92 1x2 - 2521x2 - 362
x2 - x - 6 x2 - 8x + 7 25
43. f1x2 = 44. f1x2 =
x2 - 3x - 10 x2 + 7x - 8
x2 - 49x x2 + x - 2
45. f1x2 = 46. f1x2 = with the graph and intercepts shown in the figure.
x3 - 49x x3 - 3x2 + 2x
f (x)
C For each rational function in Problems 47–52,
5
(A) Find any intercepts for the graph.
(B) Find any vertical and horizontal asymptotes for the graph.
(C) Sketch any asymptotes as dashed lines. Then sketch a graph x
25 5
of f.
(D) Graph the function in a standard viewing window using a
graphing calculator. 25
2x2 3x2
47. f1x2 = 48. f1x2 =
x2 - x - 6 x2 + x - 6
6 - 2x2 3 - 3x2
Applications
49. f1x2 = 50. f1x2 =
x2 - 9 x2 - 4 57. Average cost. A company manufacturing snowboards has
fixed costs of $200 per day and total costs of $3,800 per day
- 4x + 24 5x - 10
51. f1x2 = 52. f1x2 = 2 at a daily output of 20 boards.
x2 + x - 6 x + x - 12
(A) Assuming that the total cost per day, C1x2, is linearly
related to the total output per day, x, write an equation
with the graph and intercepts shown in the figure.
for the cost function.
f (x)
(B) The average cost per board for an output of x boards
5 is given by C1x2 = C1x2 >x. Find the average cost
function.
(C) Sketch a graph of the average cost function, including
x
25 5 any asymptotes, for 1 … x … 30.
(D) What does the average cost per board tend to as produc-
tion increases?
25
58. Average cost. A company manufacturing surfboards has 61. Minimum average cost. A consulting firm, using statistical
fixed costs of $300 per day and total costs of $5,100 per day methods, provided a veterinary clinic with the cost equation
at a daily output of 20 boards.
C1x2 = 0.000481x - 5002 3 + 60,000
(A) Assuming that the total cost per day, C1x2, is linearly
100 … x … 1,000
related to the total output per day, x, write an equation
for the cost function. where C1x2 is the cost in dollars for handling x cas-
es per month. The average cost per case is given by
(B) The average cost per board for an output of x boards is C1x2 = C1x2 >x.
given by C1x2 = C1x2 >x. Find the average cost function.
(A) Write the equation for the average cost function C.
(C) Sketch a graph of the average cost function, including
(B) Graph C on a graphing calculator.
any asymptotes, for 1 … x … 30.
(C) Use an appropriate command to find the monthly case-
(D) What does the average cost per board tend to as produc-
load for the minimum average cost per case. What is the
tion increases?
minimum average cost per case?
59. Replacement time. An office copier has an initial price of
62. Minimum average cost. The financial department of a hos-
$2,500. A service contract costs $200 for the first year and
pital, using statistical methods, arrived at the cost equation
increases $50 per year thereafter. It can be shown that the
total cost of the copier after n years is given by C1x2 = 20x3 - 360x2 + 2,300x - 1,000
C1n2 = 2,500 + 175n + 25n2 1 … x … 12
The average cost per year for n years is given by where C1x2 is the cost in thousands of dollars for handling x
C1n2 = C1n2 >n. thousand cases per month. The average cost per case is given
(A) Find the rational function C. by C1x2 = C1x2 >x.
(A) Write the equation for the average cost function C.
(B) Sketch a graph of C for 2 … n … 20.
(B) Graph C on a graphing calculator.
(C) When is the average cost per year at a minimum, and
what is the minimum average annual cost? [Hint: Refer (C) Use an appropriate command to find the monthly case-
to the sketch in part (B) and evaluate C1n2 at appropri- load for the minimum average cost per case. What is the
ate integer values until a minimum value is found.] The minimum average cost per case to the nearest dollar?
time when the average cost is minimum is frequently
63. Diet. Table 3 shows the per capita consumption of ice cream
referred to as the replacement time for the piece of
in the United States for selected years since 1987.
equipment.
(A) Let x represent the number of years since 1980 and find
(D) Graph the average cost function C on a graphing calcula- a cubic regression polynomial for the per capita con-
tor and use an appropriate command to find when the sumption of ice cream.
average annual cost is at a minimum.
(B) Use the polynomial model from part (A) to estimate (to
60. Minimum average cost. Financial analysts in a company
the nearest tenth of a pound) the per capita consumption
that manufactures DVD players arrived at the following daily
of ice cream in 2023.
cost equation for manufacturing x DVD players per day:
C1x2 = x2 + 2x + 2,000 Table 3 Per Capita Consumption
The average cost per unit at a production level of x players of Ice Cream
per day is C1x2 = C1x2 >x. Year Ice Cream (pounds)
(A) Find the rational function C. 1987 18.0
(B) Sketch a graph of C for 5 … x … 150. 1992 15.8
1997 15.7
(C) For what daily production level (to the nearest integer) 2002 16.4
is the average cost per unit at a minimum, and what is 2007 14.9
the minimum average cost per player (to the nearest 2012 13.4
cent)? [Hint: Refer to the sketch in part (B) and evalu- 2014 12.8
ate C1x2 at appropriate integer values until a minimum
Source: U.S. Department of Agriculture
value is found.]
(D) Graph the average cost function C on a graphing 64. Diet. Refer to Table 4.
calculator and use an appropriate command to find the (A) Let x represent the number of years since 2000 and
daily production level (to the nearest integer) at which find a cubic regression polynomial for the per capita
the average cost per player is at a minimum. What is the consumption of eggs.
minimum average cost to the nearest cent?
(B) Use the polynomial model from part (A) to estimate (to
the nearest integer) the per capita consumption of eggs
in 2023.
Table 4 Per Capita Consumption of Eggs Table 5 Marriages and Divorces

(per 1,000 population)
Year Number of Eggs
2002 255 Date Marriages Divorces
2004 257 1960 8.5 2.2
2006 258 1970 10.6 3.5
2008 247 1980 10.6 5.2
2010 243 1990 9.8 4.7
2012 254 2000 8.5 4.1
2014 263 2010 6.8 3.6
Source: U.S. Department of Agriculture Source: National Center for Health Statistics
65. Physiology. In a study on the speed of muscle contrac- 68. Divorce. Refer to Table 5.
tion in frogs under various loads, researchers W. O.
Fems and J. Marsh found that the speed of contraction
a cubic regression polynomial for the divorce rate.
decreases with increasing loads. In particular, they found
that the relationship between speed of contraction v (in (B) Use the polynomial model from part (A) to estimate the
centimeters per second) and load x (in grams) is given divorce rate (to one decimal place) for 2025.
approximately by
26 + 0.06x
v1x2 = x Ú 5
x 1.
(A) What does v1x2 approach as x increases?

(B) Sketch a graph of function v.
66. Learning theory. In 1917, L. L. Thurstone, a pioneer in
quantitative learning theory, proposed the rational function 2. (A) Domain: all real numbers except - 3 and 3
1
(B) x intercept: - 1; y intercept: -
a1x + c2 3
f1x2 = (C) Vertical asymptotes: x = - 3 and x = 3;
1x + c2 + b
(D) Horizontal asymptote: y = 0
to model the number of successful acts per unit time that a
person could accomplish after x practice sessions. Suppose (E) g(x)
that for a particular person enrolled in a typing class, 10
551x + 12
f1x2 = x Ú 0
1x + 82 x
210 10
where f1x2 is the number of words per minute the person is
able to type after x weeks of lessons.
210
(A) What does f1x2 approach as x increases?
(B) Sketch a graph of function f, including any vertical or 3. Vertical asymptote: x = - 5
horizontal asymptotes. Horizontal asymptote: none
67. Marriage. Table 5 shows the marriage and divorce rates per 4. No vertical asymptotes for t Ú 0;
1,000 population for selected years since 1960. y = 25 is a horizontal asymptote. N(t)
N1t2 approaches 25 as t increases 30
without bound. It appears that 25 com- 20
a cubic regression polynomial for the marriage rate.
ponents per day would be the upper 10
(B) Use the polynomial model from part (A) to estimate the limit that an employee would t
marriage rate (to one decimal place) for 2025. be expected to assemble. 0 50 100

■■ Exponential Functions This section introduces an important class of functions called exponential functions.
■■ Base e Exponential Function
These functions are used extensively in modeling and solving a wide variety of real-
world problems, including growth of money at compound interest, growth of popula-
■■ Growth and Decay Applications tions, radioactive decay, and learning associated with the mastery of such devices as a
■■ Compound Interest new computer or an assembly process in a manufacturing plant.
Exponential Functions
We start by noting that
f1x2 = 2x and g1x2 = x2
are not the same function. Whether a variable appears as an exponent with a constant
base or as a base with a constant exponent makes a big difference. The function g is a
quadratic function, which we have already discussed. The function f is a new type of
function called an exponential function. In general,
DEFINITION Exponential Function

The equation
f1x2 = bx b 7 0, b ∙ 1
defines an exponential function for each different constant b, called the base. The
domain of f is the set of all real numbers, and the range of f is the set of all positive
real numbers.
y We require the base b to be positive to avoid imaginary numbers such as

10 1 - 22 1>2 = 1 - 2 = i 12. We exclude b = 1 as a base, since f1x2 = 1x = 1 is a
constant function, which we have already considered.
When asked to hand-sketch graphs of equations such as y = 2x or y = 2 - x,
many students do not hesitate. [Note: 2 - x = 1>2x = 11>22 x.] They make tables
5 by assigning integers to x, plot the resulting points, and then join these points with a
smooth curve as in Figure 1. The only catch is that we have not defined 2x for all real
numbers. From Appendix B, Section B.6, we know what 25, 2 -3, 22>3, 2 - 3>5, 21.4, and
2 - 3.14 mean (that is, 2p, where p is a rational number), but what does
x
25 5 2 22
x
Figure 1 y ∙ 2
mean? The question is not easy to answer at this time. In fact, a precise definition of
2 12 must wait for more advanced courses, where it is shown that
2x
names a positive real number for x any real number, and that the graph of y = 2x is
as indicated in Figure 1.
It is useful to compare the graphs of y = 2x and y = 2 - x by plotting both on the
same set of coordinate axes, as shown in Figure 2A. The graph of
f1x2 = bx b 7 1 (Fig. 2B)
looks very much like the graph of y = 2x, and the graph of
f1x2 = bx 0 6 b 6 1 (Fig. 2B)
looks very much like the graph of y = 2 - x. Note that in both cases the x axis is a
horizontal asymptote for the graphs.
SECTION 1.5 Exponential Functions 75
y y
10
y 5 (2
1 x y 5 bx y 5 bx
2) 52
2x
5 y 5 2x
0,b,1 b.1
x x
25 5 Domain 5 (2`, `) Range 5 (0, `)
(A) (B)
Figure 2 Exponential functions
The graphs in Figure 2 suggest the following general properties of exponential

functions, which we state without proof:
THEOREM 1 Basic Properties of the Graph of f1x2 = b x , b 7 0, b ∙ 1

1. All graphs will pass through the point 10, 12. b0 = 1 for any
2. All graphs are continuous curves, with no holes or jumps. permissible base b.
3. The x axis is a horizontal asymptote.
4. If b 7 1, then bx increases as x increases.
5. If 0 6 b 6 1, then bx decreases as x increases.
CONCEPTUAL INSIGHT
Recall that the graph of a rational function has at most one horizontal asymptote
and that it approaches the horizontal asymptote (if one exists) both as x S ∞ and
as x S - ∞ (see Section 1.4). The graph of an exponential function, on the other
hand, approaches its horizontal asymptote as x S ∞ or as x S - ∞, but not both.
In particular, there is no rational function that has the same graph as an exponen-
tial function.
The use of a calculator with the key y x , or its equivalent, makes the graphing of
exponential functions almost routine. Example 1 illustrates the process.
EXAMPLE 1 Graphing Exponential Functions Sketch a graph of y = 1 12 2 4x, - 2 … x … 2.

SOLUTION Use a calculator to create the table of values shown. Plot these points, and
then join them with a smooth curve as in Figure 3. y
10
x y
-2 0.031
-1 0.125
0 0.50
5
1 2.00
2 8.00
x
22 2
Figure 3 Graph of y ∙ 1 12 2 4 x
Matched Problem 1 Sketch a graph of y = 1 12 2 4 - x, - 2 … x … 2.
Exponential functions, whose domains include irrational numbers, obey the famil-
iar laws of exponents discussed in Appendix B, Section B.6 for rational exponents. We
summarize these exponent laws here and add two other important and useful properties.
THEOREM 2 Properties of Exponential Functions

For a and b positive, a ∙ 1, b ∙ 1, and x and y real,
1. Exponent laws:
Reminder ax 42y
axay = a x + y = ax - y = 42y - 5y = 4 - 3y
1 -22 2 = 22, but this equation ay 45y
does not contradict property 3 of a x ax
Theorem 2. In Theorem 2, both a 1ax 2 y = axy 1ab2 x = axbx a b = x
b b
and b must be positive.
2. ax = ay if and only if x = y If 75t + 1 = 73t - 3, then
5t + 1 = 3t - 3, and t = - 2.
3. For x ∙ 0,
ax = bx if and only if a = b If a5 = 25, then a = 2.
Base e Exponential Function

Of all the possible bases b we can use for the exponential function y = bx, which
ones are the most useful? If you look at the keys on a calculator, you will probably see
10 x and ex . It is clear why base 10 would be important, because our number system is
a base 10 system. But what is e, and why is it included as a base? It turns out that base e
is used more frequently than all other bases combined. The reason for this is that certain
formulas and the results of certain processes found in calculus and more advanced math-
ematics take on their simplest form if this base is used. This is why you will see e used
extensively in expressions and formulas that model real-world phenomena. In fact, its use
is so prevalent that you will often hear people refer to y = ex as the exponential function.
The base e is an irrational number and, like p, it cannot be represented exactly by
any finite decimal or fraction. However, e can be approximated as closely as we like
by evaluating the expression
1 x
a1 + b (1)
x
for sufficiently large values of x. What happens to the value of expression (1) as x
increases without bound? Think about this for a moment before proceeding. Maybe
you guessed that the value approaches 1, because
1
1 +
x
approaches 1, and 1 raised to any power is 1. Let us see if this reasoning is correct by
actually calculating the value of the expression for larger and larger values of x. Table 1
summarizes the results.
Table 1
1 x
x a1 + b
x
1 2
10 2.593 74…
100 2.704 81…
1,000 2.716 92…
10,000 2.718 14…
100,000 2.718 26…
1,000,000 2.718 28…
Interestingly, the value of expression (1) is never close to 1 but seems to be ap-
proaching a number close to 2.7183. In fact, as x increases without bound, the value
of expression (1) approaches an irrational number that we call e. The irrational num-
ber e to 12 decimal places is
y
y 5 e2x y 5 ex e ∙ 2.718 281 828 459
10
Compare this value of e with the value of e1 from a calculator.
DEFINITION Exponential Functions with Base e and Base 1>e

5
The exponential functions with base e and base 1>e, respectively, are defined by
y = ex and y = e - x
Domain: 1 - ∞, ∞ 2
x
25 5 Range: 10, ∞ 2 (see Fig. 4)
Figure 4

Graph the functions f1x2 = ex, g1x2 = 2x, and h1x2 = 3x on the same set of co-
ordinate axes. At which values of x do the graphs intersect? For positive values of x,
which of the three graphs lies above the other two? Below the other two? How does
your answer change for negative values of x?
Growth and Decay Applications

Functions of the form y = cekt, where c and k are constants and the independent
variable t represents time, are often used to model population growth and radioactive
decay. Note that if t = 0, then y = c. So the constant c represents the initial popula-
tion (or initial amount). The constant k is called the relative growth rate and has the
following interpretation: Suppose that y = cekt models the population of a country,
where y is the number of persons and t is time in years. If the relative growth rate is
k = 0.02, then at any time t, the population is growing at a rate of 0.02y persons (that
is, 2% of the population) per year.
We say that population is growing continuously at relative growth rate k to
mean that the population y is given by the model y = cekt.
EXAMPLE 2 Exponential Growth Cholera, an intestinal disease, is caused by a cholera bac-

terium that multiplies exponentially. The number of bacteria grows continuously at
relative growth rate 1.386, that is,
N
N = N0 e1.386t
10,000
where N is the number of bacteria present after t hours and N0 is the number of bac-
teria present at the start 1t = 02. If we start with 25 bacteria, how many bacteria
(to the nearest unit) will be present
(A) In 0.6 hour? (B) In 3.5 hours?
SOLUTION Substituting N0 = 25 into the preceding equation, we obtain
N = 25e1.386t The graph is shown in Figure 5.

t (A) Solve for N when t = 0.6:
0 5
Time (hours)
N = 25e1.38610.62 Use a calculator.
Figure 5
= 57 bacteria
(B) Solve for N when t = 3.5:

N = 25e1.38613.52 Use a calculator.
= 3,197 bacteria
Matched Problem 2 Refer to the exponential growth model for cholera in
Example 2. If we start with 55 bacteria, how many bacteria (to the nearest unit)
will be present
(A) In 0.85 hour? (B) In 7.25 hours?
EXAMPLE 3 Exponential Decay Cosmic-ray bombardment of the atmosphere produces neu-

trons, which in turn react with nitrogen to produce radioactive carbon-14 1 14C2.
Radioactive 14C enters all living tissues through carbon dioxide, which is first ab-
sorbed by plants. As long as a plant or animal is alive, 14C is maintained in the living
organism at a constant level. Once the organism dies, however, 14C decays accord-
ing to the equation
A = A 0e - 0.000124t
where A is the amount present after t years and A0 is the amount present at time
t = 0.
(A) If 500 milligrams of 14C is present in a sample from a skull at the time of death,
how many milligrams will be present in the sample in 15,000 years? Compute
A the answer to two decimal places.
500 (B) The half-life of 14C is the time t at which the amount present is one-half the
amount at time t = 0. Use Figure 6 to estimate the half-life of 14C.
400
SOLUTION Substituting A0 = 500 in the decay equation, we have
Milligrams
300
A = 500e - 0.000124t See the graph in Figure 6.
200
(A) Solve for A when t = 15,000:
100
A = 500e - 0.000124115,0002 Use a calculator.
t
0 50,000 = 77.84 milligrams
Years
Figure 6 (B) Refer to Figure 6, and estimate the time t at which the amount A has fallen to 250
milligrams: t ≈ 6,000 years. (Finding the intersection of y1 = 500e - 0.000124x
and y2 = 250 on a graphing calculator gives a better estimate: t ≈ 5,590 years.)
Matched Problem 3 Refer to the exponential decay model in Example 3.

14
How many milligrams of C would have to be present at the beginning in order to
have 25 milligrams present after 18,000 years? Compute the answer to the nearest
milligram.
If you buy a new car, it is likely to depreciate in value by several thousand dollars
during the first year you own it. You would expect the value of the car to decrease in
each subsequent year, but not by as much as in the previous year. If you drive the car
long enough, its resale value will get close to zero. An exponential decay function
will often be a good model of depreciation; a linear or quadratic function would not
be suitable (why?). We can use exponential regression on a graphing calculator to
find the function of the form y = ab x that best fits a data set.
EXAMPLE 4 Depreciation Table 2 gives the market value of a hybrid sedan (in dollars) x years
after its purchase. Find an exponential regression model of the form y = abx for
this data set. Estimate the purchase price of the hybrid. Estimate the value of the
hybrid 10 years after its purchase. Round answers to the nearest dollar.
Table 2
x Value ($)
1 12,575
2 9,455
3 8,115
4 6,845
5 5,225
6 4,485
SOLUTION Enter the data into a graphing calculator (Fig. 7A) and find the exponen-
tial regression equation (Fig. 7B). The estimated purchase price is y1 102 =$14,910.
The data set and the regression equation are graphed in Figure 7C. Using trace, we
see that the estimated value after 10 years is $1,959.
20,000
0
0 11
(A) (B) (C)

Figure 7
Matched Problem 4 Table 3 gives the market value of a midsize sedan (in
dollars) x years after its purchase. Find an exponential regression model of the form
y = ab x for this data set. Estimate the purchase price of the sedan. Estimate the
value of the sedan 10 years after its purchase. Round answers to the nearest dollar.
Table 3
x Value ($)
1 23,125
2 19,050
3 15,625
4 11,875
5 9,450
6 7,125
Compound Interest
The fee paid to use another’s money is called interest. It is usually computed as a
percent (called interest rate) of the principal over a given period of time. If, at the
end of a payment period, the interest due is reinvested at the same rate, then the
interest earned as well as the principal will earn interest during the next payment
period. Interest paid on interest reinvested is called compound interest and may be
calculated using the following compound interest formula:
If a principal P (present value) is invested at an annual rate r (expressed as

a decimal) compounded m times a year, then the amount A (future value) in the
account at the end of t years is given by
r mt
A = P a1 + b Compound interest formula
m
For given r and m, the amount A is equal to the principal P multiplied by the expo-
nential function bt, where b = 11 + r>m2 m.
EXAMPLE 5 Compound Growth If $1,000 is invested in an account paying 10% compounded

monthly, how much will be in the account at the end of 10 years? Compute the
answer to the nearest cent.
SOLUTION We use the compound interest formula as follows:
r mt
A = P a1 + b
A m
$10,000 0.10 11221102

= 1,000 a1 + b Use a calculator.
12
= $2,707.04
$5,000
The graph of
0.10 12t
A = 1,000 a1 + b
12
t
0 10 20
Years
for 0 … t … 20 is shown in Figure 8.
Figure 8
Matched Problem 5 If you deposit $5,000 in an account paying 9% com-
pounded daily, how much will you have in the account in 5 years? Compute the
answer to the nearest cent.

Suppose that $1,000 is deposited in a savings account at an annual rate of 5%.
Guess the amount in the account at the end of 1 year if interest is compounded (1)
quarterly, (2) monthly, (3) daily, (4) hourly. Use the compound interest formula to
compute the amounts at the end of 1 year to the nearest cent. Discuss the accuracy
of your initial guesses.
Explore and Discuss 2 suggests that if $1,000 were deposited in a savings

account at an annual interest rate of 5%, then the amount at the end of 1 year would
be less than $1,051.28, even if interest were compounded every minute or every sec-
ond. The limiting value, approximately $1,051.271 096, is said to be the amount in
the account if interest were compounded continuously.
If a principal, P, is invested at an annual rate, r, and compounded continuously,
then the amount in the account at the end of t years is given by
A = Pert Continuous compound interest formula
where the constant e ≈ 2.718 28 is the base of the exponential function.
EXAMPLE 6 Continuous Compound Interest If $1,000 is invested in an account paying 10%

compounded continuously, how much will be in the account at the end of 10 years?
Compute the answer to the nearest cent.
SOLUTION We use the continuous compound interest formula:
A = Pert = 1000e0.101102 = 1000e = $2,718.28

Compare with the answer to Example 5.
Matched Problem 6 If you deposit $5,000 in an account paying 9%

compounded continuously, how much will you have in the account in 5 years?
Compute the answer to the nearest cent.
The formulas for compound interest and continuous compound interest are summa-
rized below for convenient reference.
SUMMARY
r mt
Compound Interest: A = Pa1 + b
m
Continuous Compound Interest: A = Pert
where A = amount (future value) at the end of t years
P = principal (present value)
r = annual rate (expressed as a decimal)
m = number of compounding periods per year
t = time in years
Exercises 1.5
A 1. Match each equation with the graph of f, g, h, or k in the y
figure. f g h k
(A) y = 2x (B) y = 10.22 x

5
(C) y = 4x (D) y = 1 13 2 x
y
f g h k
5 x
25 5
Graph each function in Problems 3–10 over the indicated interval.

3. y = 5x; 3 - 2, 24 4. y = 3x; 3 - 3, 34
x
25 5 5. y = 1 15 2 x = 5 - x; 3 - 2, 24 6. y = 1 13 2 x = 3 - x; 3 - 3, 34
7. f1x2 = - 5x; 3 - 2, 24 8. g1x2 = - 3 - x; 3 - 3, 34
2. Match each equation with the graph of f, g, h, or k in the
figure. 9. y = - e - x; 3 - 3, 34 10. y = - ex; 3 - 3, 34
x
(A) y = 1 14 2 (B) y = 10.52 x
(C) y = 5x (D) y = 3x
B In Problems 11–18, describe verbally the transformations that C In Problems 39–46, solve each equation for x. (Remember:
can be used to obtain the graph of g from the graph of f (see ex ∙ 0 and e - x ∙ 0 for all values of x).
Section 1.2).
39. xe - x + 7e - x = 0 40. 10xex - 5ex = 0
x x
11. g1x2 = - 2 ; f1x2 = 2
41. 2x2ex - 8ex = 0 42. 3x2e-x - 48e- x = 0
x-2 x
12. g1x2 = 2 ; f1x2 = 2
43. e4x - e = 0 44. e4x + e = 0
x+1 x
13. g1x2 = 3 ; f1x2 = 3
45. e3x - 1 + e = 0 46. e3x - 1 - e = 0
14. g1x2 = - 3 ; f1x2 = 3xx
Graph each function in Problems 47–50 over the indicated interval.
15. g1x2 = ex + 1; f1x2 = ex
47. h1x2 = x12x 2; 3 - 5, 04 48. m1x2 = x13 - x 2; 30, 34
16. g1x2 = ex - 2; f1x2 = ex
17. g1x2 = 2e - 1x + 22; f1x2 = e - x 100 300
49. N = ; 30, 54 50. N = ; 30, 54
1 + e-t 1 + 4e-t
18. g1x2 = 0.25e - 1x - 42; f1x2 = e - x
19. Use the graph of f shown in the figure to sketch the graph of
each of the following.
Applications
(A) y = f1x2 - 1 (B) y = f1x + 22
In all problems involving days, a 365-day year is assumed.
(C) y = 3f1x2 - 2 (D) y = 2 - f1x - 32
y
51. Continuous compound interest. Find the value of an
investment of $100,000 in 9 years if it earns an annual rate of
5 2.85% compounded continuously.
52. Continuous compound interest. Find the value of an invest-
ment of $24,000 in 7 years if it earns an annual rate of 4.35%
x compounded continuously.
25 5
53. Compound growth. Suppose that $2,500 is invested at 7%

compounded quarterly. How much money will be in the
25 account in
Figure for 19 and 20 (A) 3
year? (B) 15 years?
4
20. Use the graph of f shown in the figure to sketch the graph of Compute answers to the nearest cent.
each of the following.
54. Compound growth. Suppose that $4,000 is invested at 6%
(A) y = f1x2 + 2 (B) y = f1x - 32 compounded weekly. How much money will be in the ac-
(C) y = 2 f1x2 - 4 (D) y = 4 - f1x + 22 count in
1
In Problems 21–26, graph each function over the indicated interval. (A) 2 year? (B) 10 years?
21. f1t2 = 2t>10; 3 - 30, 304 22. G1t2 = 3t>100; 3- 200, 2004 Compute answers to the nearest cent.
55. Finance. A person wishes to have $15,000 cash for a new
23. y = - 3 + e1 + x; 3 - 4, 24 24. N = 1 + ex - 3; 3 - 2, 64 car 5 years from now. How much should be placed in an
25. y = e∙x∙; 3 - 3, 34 26. y = e-∙x∙; 3 - 3, 34 account now, if the account pays 6.75% compounded
weekly? Compute the answer to the nearest dollar.
27. Find all real numbers a such that a2 = a - 2. Explain why this
does not violate the second exponential function property in 56. Finance. A couple just had a baby. How much should they
Theorem 2 on page 76. invest now at 5.5% compounded daily in order to have
$40,000 for the child’s education 17 years from now?
28. Find real numbers a and b such that a ∙ b but a4 = b4. Ex-
Compute the answer to the nearest dollar.
plain why this does not violate the third exponential function
property in Theorem 2 on page 76. 57. Money growth. Commonwealth Bank of Australia offers
term deposits with different interest payment options. The
In Problems 29–38, solve each equation for x. rates for 24-month term deposits are the following:
29. 22x + 5 = 2101 30. 3x + 4 = 32x - 5 (A) 2.40% compounded four weekly
2
x 4x + 21 x2 - x 42
31. 9 = 9 32. 5 = 5 (B) 2.50% compounded six monthly
5 5
33. 13x + 92 = 32x 34. 15x + 62 7 = 1117 (C) 2.60% compounded annually
2 2 2 2
35. 1x + 52 = 12x - 142 36. 12x + 12 = 13x - 12 Compute the value of $10,000 invested in each interest pay-
37. 15x + 182 = 1x + 62 4 4
38. 18x - 112 = 17x - 1926 6 ment option at the end of 2 years.
58. Money growth. Bankwest, a division of Commonwealth Table 5 Mobile Data Traffic
Bank of Australia, offers online term deposits with different (exabytes per month)
interest payment options. The rates for 60-month term depos- Year Traffic
its are the following:
2015 3.7
(A) 2.62% compounded monthly 2016 6.2
(B) 2.62% compounded quarterly 2017 9.9
2018 14.9
(C) 2.63% compounded semiannually 2019 21.7
(D) 2.65% compounded annually 2020 30.6
Source: Cisco Systems Inc.
Compute the value of $10,000 invested in each interest pay-
ment option at the end of 5 years.
63. Marine biology. Marine life depends on the microscopic
59. Advertising. A company is trying to introduce a new prod- plant life that exists in the photic zone, a zone that goes to
uct to as many people as possible through television adver- a depth where only 1% of surface light remains. In some
tising in a large metropolitan area with 2 million possible waters with a great deal of sediment, the photic zone may go
viewers. A model for the number of people N (in millions) down only 15 to 20 feet. In some murky harbors, the intensity
who are aware of the product after t days of advertising was of light d feet below the surface is given approximately by
found to be
I = I0e - 0.23d
N = 2 1 1 - e-0.037t 2
where I0 is the intensity of light at the surface. What percent-
Graph this function for 0 … t … 50. What value does N ap- age of the surface light will reach a depth of
proach as t increases without bound?
(A) 10 feet? (B) 20 feet?
60. Learning curve. People assigned to assemble circuit boards
for a computer manufacturing company undergo on-the-job 64. Marine biology. Refer to Problem 63. Light intensity I rela-
training. From past experience, the learning curve for the tive to depth d (in feet) for one of the clearest bodies of water
average employee is given by in the world, the Sargasso Sea, can be approximated by
N = 4011 - e - 0.12t 2 I = I0e - 0.00942d
where N is the number of boards assembled per day after where I0 is the intensity of light at the surface. What percent-
t days of training. Graph this function for 0 … t … 30. What age of the surface light will reach a depth of
is the maximum number of boards an average employee can (A) 50 feet? (B) 100 feet?
be expected to produce in 1 day?
65. Population growth. In 2015, the estimated population of South
61. Internet users. Table 4 shows the number of individuals Sudan was 12 million with a relative growth rate of 4.02%.
worldwide who could access the internet from home for
selected years since 2000. (A) Write an equation that models the population growth in
South Sudan, letting 2015 be year 0.
an exponential regression model 1y = abx 2 for the num- (B) Based on the model, what is the expected population of
ber of internet users. South Sudan in 2025?
(B) Use the model to estimate the number of internet users 66. Population growth. In 2015, the estimated population of
in 2024. Brazil was 204 million with a relative growth rate of 0.77%.
(A) Write an equation that models the population growth in
Table 4 Internet Users (billions) Brazil, letting 2015 be year 0.
Year Users (B) Based on the model, what is the expected population of
2000 0.41 Brazil in 2030?
2004 0.91
67. Population growth. In 2015, the estimated population of
2008 1.58
Japan was 127 million with a relative growth rate of - 0.16%.
2012 2.02
2016 3.42 (A) Write an equation that models the population growth in
Source: Internet Stats Live Japan, letting 2015 be year 0.
(B) Based on the model, what is the expected population in
62. Mobile data traffic. Table 5 shows estimates of mobile data Japan in 2030?
traffic, in exabytes 11018 bytes2 per month, for years from
2015 to 2020. 68. World population growth. From the dawn of humanity to
1830, world population grew to one billion people. In 100
(A) Let x represent the number of years since 2015 and find more years (by 1930) it grew to two billion, and 3 billion
an exponential regression model 1y = abx 2 for mobile more were added in only 60 years (by 1990). In 2016, the
data traffic. estimated world population was 7.4 billion with a relative
(B) Use the model to estimate the mobile data traffic in growth rate of 1.13%.
2025.
(A) Write an equation that models the world population 2. (A) 179 bacteria (B) 1,271,659 bacteria
growth, letting 2016 be year 0. 3. 233 mg
(B) Based on the model, what is the expected world popula- 4. Purchase price: $30,363; value after 10 yr: $2,864
tion (to the nearest hundred million) in 2025? In 2033?
1. y
10 5. $7,841.13
6. $7,841.56
x
22 2
1.6 Logarithmic Functions

■■ Inverse Functions Find the exponential function keys 10x and ex on your calculator. Close to these
■■ Logarithmic Functions
keys you will find the log and ln keys. The latter two keys represent logarithmic
functions, and each is closely related to its nearby exponential function. In fact, the ex-
■■ Properties of Logarithmic Functions ponential function and the corresponding logarithmic function are said to be inverses
■■ Calculator Evaluation of Logarithms of each other. In this section we will develop the concept of inverse functions and use
it to define a logarithmic function as the inverse of an exponential function. We will
■■ Applications
then investigate basic properties of logarithmic functions, use a calculator to evaluate
them for particular values of x, and apply them to real-world problems.
Logarithmic functions are used in modeling and solving many types of prob-
lems. For example, the decibel scale is a logarithmic scale used to measure sound
intensity, and the Richter scale is a logarithmic scale used to measure the force of an
earthquake. An important business application has to do with finding the time it takes
money to double if it is invested at a certain rate compounded a given number of
times a year or compounded continuously. This requires the solution of an exponen-
tial equation, and logarithms play a central role in the process.
Inverse Functions
x ∙x∙
Look at the graphs of f1x2 = and g1x2 = in Figure 1:
2 2
f (x) g(x)
5 5
x x
25 5 25 5
25 25
(A) f (x) 5
x
(B) g(x) 5
|x|
2 2
Figure 1
SECTION 1.6 Logarithmic Functions 85
Because both f and g are functions, each domain value corresponds to exactly
one range value. For which function does each range value correspond to exactly one
domain value? This is the case only for function f. Note that for function f, the range
value 2 corresponds to the domain value 4. For function g the range value 2 corre-
sponds to both - 4 and 4. Function f is said to be one-to-one.
DEFINITION One-to-One Functions

Reminder A function f is said to be one-to-one if each range value corresponds to exactly one
domain value.
We say that the function f is
increasing on an interval 1a, b2
if f1x2 2 7 f1x1 2 whenever It can be shown that any continuous function that is either increasing or decreas-
a 6 x1 6 x2 6 b and f is decreasing ing for all domain values is one-to-one. If a continuous function increases for some
on 1a, b2 if f1x2 2 6 f1x1 2 whenever domain values and decreases for others, then it cannot be one-to-one. Figure 1 shows
a 6 x1 6 x2 6 b. an example of each case.
Graph f1x2 = 2x and g1x2 = x2. For a range value of 4, what are the correspond-
ing domain values for each function? Which of the two functions is one-to-one?
Explain why.
Starting with a one-to-one function f, we can obtain a new function called the
inverse of f.
DEFINITION Inverse of a Function

If f is a one-to-one function, then the inverse of f is the function formed by inter-
changing the independent and dependent variables for f. Thus, if 1a, b2 is a point on
the graph of f, then 1b, a2 is a point on the graph of the inverse of f.
Note: If f is not one-to-one, then f does not have an inverse.
In this course, we are interested in the inverses of exponential functions, called

logarithmic functions.
Logarithmic Functions
If we start with the exponential function f defined by
y = 2x (1)
and interchange the variables, we obtain the inverse of f:
x = 2y (2)
We call the inverse the logarithmic function with base 2, and write
y = log2 x if and only if x = 2y
We can graph y = log2 x by graphing x = 2y since they are equivalent. Any ordered
pair of numbers on the graph of the exponential function will be on the graph of the
logarithmic function if we interchange the order of the components. For example,
13, 82 satisfies equation (1) and 18, 32 satisfies equation (2). The graphs of y = 2x
and y = log2 x are shown in Figure 2. Note that if we fold the paper along the dashed
line y = x in Figure 2, the two graphs match exactly. The line y = x is a line of sym-
metry for the two graphs.
Exponential Logarithmic
y
y5 2x Function Function
y5x x y ∙ 2x x ∙ 2y y
10
-3 1 1 -3
8 8
-2 1 1 -2
5 x 5 2y 4 4
or -1 1 1 -1
y 5 log2 x 2 2
x 0 1 1 0
25 5 10
1 2 2 1
2 4 4 2
25
3 8 8 3
Figure 2
Ordered
"
"
pairs
reversed
In general, since the graphs of all exponential functions of the form

f1x2 = bx, b ∙ 1, b 7 0, are either increasing or decreasing, exponential functions
have inverses.
DEFINITION Logarithmic Functions

The inverse of an exponential function is called a logarithmic function. For
y
y 5 bx b 7 0 and b ∙ 1,
10
y5x Logarithmic form Exponential form
y = logb x is equivalent to x = by
5
y 5 logb x The log to the base b of x is the exponent to which b must be raised to obtain x. [Re-
member: A logarithm is an exponent.] The domain of the logarithmic function is the
x set of all positive real numbers, which is also the range of the corresponding exponen-
25 5 10
tial function, and the range of the logarithmic function is the set of all real numbers,
25 which is also the domain of the corresponding exponential function. Typical graphs of
an exponential function and its inverse, a logarithmic function, are shown in Figure 3.
Figure 3
CONCEPTUAL INSIGHT
Because the domain of a logarithmic function consists of the positive real num-
bers, the entire graph of a logarithmic function lies to the right of the y axis. In
contrast, the graphs of polynomial and exponential functions intersect every ver-
tical line, and the graphs of rational functions intersect all but a finite number of
vertical lines.
The following examples involve converting logarithmic forms to equivalent ex-

ponential forms, and vice versa.
EXAMPLE 1 Logarithmic–Exponential Conversions Change each logarithmic form to an

equivalent exponential form:
(A) log5 25 = 2 (B) log9 3 = 12 (C) log2 1 14 2 = - 2
SOLUTION
(A) log5 25 = 2 is equivalent to 25 = 52
(B) log9 3 = 12 is equivalent to 3 = 91>2
(C) log2 1 14 2 = - 2 is equivalent to 1
4 = 2
-2
Matched Problem 1 Change each logarithmic form to an equivalent exponen-

tial form:
(A) log3 9 = 2 (B) log4 2 = 1
2 (C) log3 1 19 2 = - 2
EXAMPLE 2 Exponential–Logarithmic Conversions Change each exponential form to an

equivalent logarithmic form:
(A) 64 = 43 (B) 6 = 136 (C) 18 = 2 -3
SOLUTION
(A) 64 = 43 is equivalent to log4 64 = 3
(B) 6 = 136 is equivalent to log36 6 = 12
(C) 18 = 2 -3 is equivalent to log2 1 18 2 = - 3
Matched Problem 2 Change each exponential form to an equivalent logarithmic

form:
(A) 49 = 72 (B) 3 = 19 (C) 13 = 3 -1
To gain a deeper understanding of logarithmic functions and their relationship to

exponential functions, we consider a few problems where we want to find x, b, or y
in y = logb x, given the other two values. All values are chosen so that the problems
can be solved exactly without a calculator.
EXAMPLE 3 Solutions of the Equation y = logb x Find y, b, or x, as indicated.

(A) Find y: y = log4 16 (B) Find x: log2 x = - 3
(C) Find b: logb 100 = 2
SOLUTION
(A) y = log4 16 is equivalent to 16 = 4y. So,
y = 2
(B) log2 x = - 3 is equivalent to x = 2 -3. So,
1 1
x = 3 =
2 8
(C) logb 100 = 2 is equivalent to 100 = b2. So,
b = 10 Recall that b cannot be negative.
Matched Problem 3 Find y, b, or x, as indicated.

(A) Find y: y = log9 27 (B) Find x: log3 x = - 1
(C) Find b: logb 1,000 = 3
Properties of Logarithmic Functions

The properties of exponential functions (Section 1.5) lead to properties of logarith-
mic functions. For example, consider the exponential property bxby = bx + y. Let
M = bx, N = by. Then
logb MN = logb 1bxby 2 = logb b x + y = x + y = logb M + logb N
So logb MN = logb M + logb N, that is, the logarithm of a product is the sum of the
logarithms. Similarly, the logarithm of a quotient is the difference of the logarithms.
These properties are among the eight useful properties of logarithms that are listed in
Theorem 1.
THEOREM 1 Properties of Logarithmic Functions

If b, M, and N are positive real numbers, b ∙ 1, and p and x are real numbers, then
1. logb 1 = 0 5. logb MN = logb M + logb N
M
2. logb b = 1 6. logb = logb M - logb N
N
3. logb bx = x 7. logb M p = p logb M
4. blogb x = x, x 7 0 8. logb M = logb N if and only if M = N
EXAMPLE 4 Using Logarithmic Properties Use logarithmic properties to write in simpler

form:
wx loge x
(A) logb (B) logb 1wx2 3>5 (C) ex loge b (D)
yz loge b
SOLUTION
wx
(A) logb = logb wx - logb yz
yz
= logb w + logb x - 1logb y + logb z2
= logb w + logb x - logb y - logb z
3
(B) logb 1wx2 3>5 = 5 logb wx = 35 1logb w + logb x2
x
(C) ex loge b = eloge b = bx
loge x loge 1blogb x 2 1logb x21loge b2
(D) = = = logb x
loge b loge b loge b
Matched Problem 4 Write in simpler form, as in Example 4.

R R 2>3 log2 x
(A) logb (B) logb a b (C) 2u log2 b (D)
ST S log2 b
The following examples and problems will give you additional practice in using basic
logarithmic properties.
EXAMPLE 5 Solving Logarithmic Equations Find x so that

3
2 log b 4 - 23 logb 8 + logb 2 = logb x
3
SOLUTION 2 log b 4 - 23 logb8 + logb 2 = logb x Use property 7.
logb 43>2 - logb82>3 + logb 2 = logb x Simplify.

logb 8 - logb 4 + logb 2 = logb x Use properties 5 and 6.
8#2
logb = logb x Simplify.
4
logb 4 = logb x Use property 8.
x = 4
Matched Problem 5 Find x so that 3 logb 2 + 12 logb 25 - logb 20 = logb x.
EXAMPLE 6 Solving Logarithmic Equations Solve: log10 x + log10 1x + 12 = log10 6.

SOLUTION log10 x + log10 1x + 12 = log10 6 Use property 5.
log10 3x1x + 124 = log10 6 Use property 8.
x1x + 12 = 6 Expand.
x2 + x - 6 = 0 Solve by factoring.
1x + 321x - 22 = 0
x = - 3, 2
We must exclude x = - 3, since the domain of the function log10 x is 10, ∞ 2; so
x = 2 is the only solution.
Matched Problem 6 Solve: log3 x + log3 1x - 32 = log3 10.
Calculator Evaluation of Logarithms

Of all possible logarithmic bases, e and 10 are used almost exclusively. Before we
can use logarithms in certain practical problems, we need to be able to approximate
the logarithm of any positive number either to base 10 or to base e. And conversely,
if we are given the logarithm of a number to base 10 or base e, we need to be able
to approximate the number. Historically, tables were used for this purpose, but now
calculators make computations faster and far more accurate.
Common logarithms are logarithms with base 10. Natural logarithms are log-
arithms with base e. Most calculators have a key labeled “log” (or “LOG”) and a key
labeled “ln” (or “LN”). The former represents a common (base 10) logarithm and the
latter a natural (base e) logarithm. In fact, “log” and “ln” are both used extensively in
mathematical literature, and whenever you see either used in this book without a base
indicated, they will be interpreted as follows:
Common logarithm: log x means log10 x
Natural logarithm: ln x means loge x
Finding the common or natural logarithm using a calculator is very easy. On
some calculators, you simply enter a number from the domain of the function and
press log or ln. On other calculators, you press either log or ln, enter a number
from the domain, and then press enter. Check the user’s manual for your calculator.
EXAMPLE 7 Calculator Evaluation of Logarithms Use a calculator to evaluate each to six

decimal places:
(A) log 3,184 (B) ln 0.000 349 (C) log 1 - 3.242
SOLUTION
(A) log 3,184 = 3.502 973
(B) ln 0.000 349 = - 7.960 439
(C) log 1 - 3.242 = Error - 3.24 is not in the domain of the log function.
Matched Problem 7 Use a calculator to evaluate each to six decimal places:

(A) log 0.013 529 (B) ln 28.693 28 (C) ln 1 - 0.4382
Given the logarithm of a number, how do you find the number? We make direct use
of the logarithmic-exponential relationships, which follow from the definition of
logarithmic function given at the beginning of this section.
log x ∙ y is equivalent to x ∙ 10 y
ln x ∙ y is equivalent to x ∙ ey
EXAMPLE 8 Solving logb x = y for x Find x to four decimal places, given the indicated loga-
rithm:
(A) log x = - 2.315 (B) ln x = 2.386
SOLUTION
(A) log x = - 2.315 Change to equivalent exponential form.
x = 10 -2.315 Evaluate with a calculator.
= 0.0048
(B) ln x = 2.386 Change to equivalent exponential form.
x = e2.386 Evaluate with a calculator.
= 10.8699
Matched Problem 8 Find x to four decimal places, given the indicated logarithm:
(A) ln x = - 5.062 (B) log x = 2.0821
We can use logarithms to solve exponential equations.
EXAMPLE 9 Solving Exponential Equations Solve for x to four decimal places:

(A) 10 x = 2 (B) e x = 3 (C) 3x = 4
SOLUTION
(A) 10 x = 2 Take common logarithms of both sides.
x
log 10 = log 2 Use property 3.
x = log 2 Use a calculator.
= 0.3010 To four decimal places
x
e = 3 Take natural logarithms of both sides.
(B) ln ex = ln 3 Use property 3.

x = ln 3 Use a calculator.
(C) 3x = 4 Take either natural or common logarithms of both sides.

(We choose common logarithms.)
log 3x = log 4 Use property 7.
x log 3 = log 4 Solve for x.
log 4
x = Use a calculator.
log 3
Matched Problem 9 Solve for x to four decimal places:

x
(A) 10 = 7 (B) ex = 6 (C) 4x = 5
Exponential equations can also be solved graphically by graphing both sides of

an equation and finding the points of intersection. Figure 4 illustrates this approach
for the equations in Example 9.
5 5 5
22 2 22 2 22 2
21 21 21
(A) y1 5 10 x (B) y1 5 e x (C) y1 5 3 x

y2 5 2 y2 5 3 y2 5 4
Figure 4 Graphical solution of exponential equations

Discuss how you could find y = log5 38.25 using either natural or common loga-
rithms on a calculator. [Hint: Start by rewriting the equation in exponential form.]
Remark In the usual notation for natural logarithms, the simplifications of

Example 4, parts (C) and (D) on page 88, become
ln x
ex ln b = b x and = logb x
ln b
With these formulas, we can change an exponential function with base b, or a log-
arithmic function with base b, to expressions involving exponential or logarithmic
functions, respectively, to the base e. Such change-of-base formulas are useful in
calculus.
Applications
A convenient and easily understood way of comparing different investments is to
use their doubling times—the length of time it takes the value of an investment to
double. Logarithm properties, as you will see in Example 10, provide us with just the
right tool for solving some doubling-time problems.
EXAMPLE 10 Doubling Time for an Investment How long (to the next whole year) will it take
money to double if it is invested at 20% compounded annually?
SOLUTION We use the compound interest formula discussed in Section 1.5:
r mt
A = P a1 + b Compound interest
m
The problem is to find t, given r = 0.20, m = 1, and A = 2P; that is,

2P = P11 + 0.22 t
2 = 1.2t Solve for t by taking the natural or
common logarithm of both sides (we choose
1.2t = 2
the natural logarithm).
ln 1.2t = ln 2
t ln 1.2 = ln 2 Use property 7.
4
ln 2
t = Use a calculator.
ln 1.2
≈ 3.8 years [Note: 1ln 22 > 1ln 1.22 ∙ ln 2 - ln 1.2]
≈ 4 years To the next whole year
0 2
0
When interest is paid at the end of 3 years, the money will not be doubled; when
Figure 5 y 1 ∙ 1.2x , y 2 ∙ 2 paid at the end of 4 years, the money will be slightly more than doubled.
Example 10 can also be solved graphically by graphing both sides of the equation
2 = 1.2t, and finding the intersection point (Fig. 5).
Matched Problem 10 How long (to the next whole year) will it take money to
triple if it is invested at 13% compounded annually?
It is interesting and instructive to graph the doubling times for various rates com-
pounded annually. We proceed as follows:
A = P11 + r2 t
2P = P11 + r2 t
2 = 11 + r2 t
11 + r2 t = 2
ln 11 + r2 t = ln 2
t
t ln 11 + r2 = ln 2
70 ln 2
t =
60 ln11 + r2
50 Figure 6 shows the graph of this equation (doubling time in years) for interest rates
40 compounded annually from 1 to 70% (expressed as decimals). Note the dramatic
Years
30 t5
ln 2 change in doubling time as rates change from 1 to 20% (from 0.01 to 0.20).
ln (1 1 r) Among increasing functions, the logarithmic functions (with bases b 7 1)
20
10
increase much more slowly for large values of x than either exponential or polyno-
mial functions. When a visual inspection of the plot of a data set indicates a slowly
r
0 0.20 0.40 0.60 increasing function, a logarithmic function often provides a good model. We use
Rate compounded annually logarithmic regression on a graphing calculator to find the function of the form
Figure 6 y = a + b ln x that best fits the data.
EXAMPLE 11 Home Ownership Rates The U.S. Census Bureau published the data in Table 1
on home ownership rates. Let x represent time in years with x = 0 representing
1900. Use logarithmic regression to find the best model of the form y = a + b ln x
for the home ownership rate y as a function of time x. Use the model to predict the
home ownership rate in the United States in 2025 (to the nearest tenth of a percent).
Table 1 Home Ownership Rates

Year Rate (%)
1950 55.0
1960 61.9
1970 62.9
1980 64.4
1990 64.2
2000 67.4
2010 66.9
SOLUTION Enter the data in a graphing calculator (Fig. 7A) and find the loga-
rithmic regression equation (Fig. 7B). The data set and the regression equation are
graphed in Figure 7C. Using trace, we predict that the home ownership rate in 2025
would be 69.8%.
80
40 130
40
(A) (B) (C)

Figure 7
Matched Problem 11 Refer to Example 11. Use the model to predict the
home ownership rate in the United States in 2030 (to the nearest tenth of a percent).
Note that in Example 11 we let x = 0 represent 1900. If we let

! CAUTION x = 0 represent 1940, for example, we would obtain a different
logarithmic regression equation. We would not let x = 0 represent 1950 (the first year
in Table 1) or any later year, because logarithmic functions are undefined at 0.
Exercises 1.6
A For Problems 1–6, rewrite in equivalent exponential form. 1
17. log2 128 18. log3
1. log3 27 = 3 2. log2 32 = 5 243
19. ln e-3 20. eln1-12
3. log10 1 = 0 4. loge 1 = 0
3 4 21. eln1-32 22. ln e-1
5. log4 8 = 2 6. log8 16 = 3
For Problems 23–28, write in simpler form, as in Example 4.
For Problems 7–12, rewrite in equivalent logarithmic form. P
23. logb 24. logb FG
7. 49 = 72 8. 36 = 62 Q
9. 8 = 43>2 10. 8 = 163>4 25. logb L5 26. logb w26
11. A = bu 12. M = bx log3 P
27. 3p log3 q 28.
log3 R
In Problems 13–22, evaluate the expression without using a B For Problems 29–38, find x, y, or b without using a calculator.
calculator.
29. log10 x = - 1 30. log10 x = 1
1
13. log10 1,000,000 14. log10 1
1,000 31. logb 64 = 3 32. logb = 2
1 25
15. log10 16. log10 10,000
100,000
1 62. (A) log 72.604 (B) log 0.033 041

33. log2 = y 34. log49 7 = y
8 (C) ln 40,257 (D) ln 0.005 926 3
35. logb 81 = - 4 36. logb 10,000 = 2
For Problems 63 and 64, find x to four decimal places.
3 4 63. (A) log x = 1.1285 (B) log x = - 2.0497
37. log4 x = 38. log27 x =
2 3
(C) ln x = 2.7763 (D) ln x = - 1.8879
In Problems 39–46, discuss the validity of each statement.
If the statement is always true, explain why. If not, give a 64. (A) log x = 2.0832 (B) log x = - 1.1577
counterexample. (C) ln x = 3.1336 (D) ln x = - 4.3281
39. Every polynomial function is one-to-one. For Problems 65–70, solve each equation to four decimal places.
40. Every polynomial function of odd degree is one-to-one. 65. 10x = 12 66. 10x = 153
41. If g is the inverse of a function f, then g is one-to-one. 67. ex = 5.432 68. ex = 0.3059
42. The graph of a one-to-one function intersects each vertical 69. 1.00512t = 3 70. 1.2345t = 6
line exactly once.
Graph Problems 71–78 using a calculator and point-by-point
43. The inverse of f1x2 = 2x is g1x2 = x>2.
plotting. Indicate increasing and decreasing intervals.
44. The inverse of f1x2 = x2 is g1x2 = 1x.
71. y = ln x 72. y = - ln x
45. If f is one-to-one, then the domain of f is equal to the range
73. y = ∙ ln x ∙ 74. y = ln ∙ x ∙
of f.
75. y = 3 ln1x + 42 76. y = 3 ln x + 4
46. If g is the inverse of a function f, then f is the inverse of g.
77. y = 4 ln x - 3 78. y = 4 ln 1x - 32
C Find x in Problems 47–54.
79. Explain why the logarithm of 1 for any permissible base is 0.
47. logb x = 23 logb 8 + 12 logb 9 - logb 6
80. Explain why 1 is not a suitable logarithmic base.
48. logb x = 23 logb 27 + 2 logb 2 - logb3
81. Let p1x2 = ln x, q1x2 = 1x, and r1x2 = x. Use a
49. logb x = 32 logb 4 - 23 logb 8 + 2 logb 2 graphing calculator to draw graphs of all three functions in
50. logb x = 1
logb 36 + 13 logb 125 - 1 the same viewing window for 1 … x … 16. Discuss what
2 2 log b 100
it means for one function to be larger than another on an
51. logb x + logb 1x - 42 = logb 21 interval, and then order the three functions from largest to
52. logb 1x + 22 + logb x = logb 24 smallest for 1 6 x … 16.
3
53. log10 1x - 12 - log10 1x + 12 = 1 82. Let p1x2 = log x, q1x2 = 2 x, and r1x2 = x. Use a
graphing calculator to draw graphs of all three functions in
54. log10 1x + 62 - log10 1x - 32 = 1 the same viewing window for 1 … x … 16. Discuss what
it means for one function to be smaller than another on an
Graph Problems 55 and 56 by converting to exponential form
interval, and then order the three functions from smallest to
first.
largest for 1 6 x … 16.
55. y = log2 1x - 22 56. y = log3 1x + 22
57. Explain how the graph of the equation in Problem 55 can be
obtained from the graph of y = log2 x using a simple trans-
formation (see Section 1.2). Applications
58. Explain how the graph of the equation in Problem 56 can be 83. Doubling time. In its first 10 years the Gabelli Growth Fund
obtained from the graph of y = log3 x using a simple trans- produced an average annual return of 21.36%. Assume that
formation (see Section 1.2). money invested in this fund continues to earn 21.36% com-
59. What are the domain and range of the function defined by pounded annually. How long (to the nearest year) will it take
y = 1 + ln1x + 12? money invested in this fund to double?
60. What are the domain and range of the function defined by 84. Doubling time. In its first 10 years the Janus Flexible
y = log 1x - 12 - 1? Income Fund produced an average annual return of 9.58%.
Assume that money invested in this fund continues to earn
For Problems 61 and 62, evaluate to five decimal places using a 9.58% compounded annually. How long (to the nearest year)
calculator. will it take money invested in this fund to double?
61. (A) log 3,527.2 (B) log 0.006 913 2 85. Investing. How many years (to two decimal places) will it
take $1,000 to grow to $1,800 if it is invested at 6% com-
(C) ln 277.63 (D) ln 0.040 883
pounded quarterly? Compounded daily?
86. Investing. How many years (to two decimal places) will it of sound just below the threshold of hearing (approximately
take $5,000 to grow to $7,500 if it is invested at 8% com- 10-16 watt per square centimeter), then
pounded semiannually? Compounded monthly? I = I010 N>10
87. Continuous compound interest. How many years (to two Show that this formula can be written in the form
decimal places) will it take an investment of $35,000 to grow to
$50,000 if it is invested at 4.75% compounded continuously? I
N = 10 log
I0
88. Continuous compound interest. How many years (to two
decimal places) will it take an investment of $17,000 to grow to 92. Sound intensity: decibels. Use the formula in Problem 91
$41,000 if it is invested at 2.95% compounded continuously? (with I0 = 10-16 W>cm2) to find the decibel ratings of the
89. Supply and demand. A cordless screwdriver is sold following sounds:
through a national chain of discount stores. A marketing (A) Whisper: 10-13 W>cm2
company established price–demand and price–supply tables
(Tables 2 and 3), where x is the number of screwdrivers (B) Normal conversation: 3.16 * 10-10 W>cm2
people are willing to buy and the store is willing to sell each (C) Heavy traffic: 10-8 W>cm2
month at a price of p dollars per screwdriver.
(D) Jet plane with afterburner: 10-1 W>cm2
(A) Find a logarithmic regression model 1y = a + b ln x2
for the data in Table 2. Estimate the demand (to the 93. Agriculture. Table 4 shows the yield (in bushels per acre)
nearest unit) at a price level of $50. and the total production (in millions of bushels) for corn in
the United States for selected years since 1950. Let x repre-
Table 2 Price–Demand sent years since 1900. Find a logarithmic regression model
x p = D1x21$2 1y = a + b ln x2 for the yield. Estimate (to the nearest
bushel per acre) the yield in 2024.
1,000 91
2,000 73
Table 4 United States Corn Production
3,000 64
4,000 56 Yield (bushels Total Production
5,000 53 Year x per acre) (million bushels)
1950 50 38 2,782
(B) Find a logarithmic regression model 1y = a + b ln x2 1960 60 56 3,479
for the data in Table 3. Estimate the supply (to the near- 1970 70 81 4,802
est unit) at a price level of $50. 1980 80 98 6,867
Table 3 Price–Supply 1990 90 116 7,802
2000 100 140 10,192
x p = S1x21$2 2010 110 153 12,447
1,000 9
2,000 26 94. Agriculture. Refer to Table 4. Find a logarithmic regression
3,000 34 model 1y = a + b ln x2 for the total production. Estimate
4,000 38 (to the nearest million) the production in 2024.
5,000 41
95. World population. If the world population is now
(C) Does a price level of $50 represent a stable condition, or 7.4 billion people and if it continues to grow at an annual rate
is the price likely to increase or decrease? Explain. of 1.1% compounded continuously, how long (to the nearest
year) would it take before there is only 1 square yard of land
90. Equilibrium point. Use the models constructed in per person? (The Earth contains approximately 1.68 * 1014
Problem 89 to find the equilibrium point. Write the square yards of land.)
equilibrium price to the nearest cent and the equilibrium
quantity to the nearest unit. 96. Archaeology: carbon-14 dating. The radioactive car-
bon-14 1 14C 2 in an organism at the time of its death decays
91. Sound intensity: decibels. Because of the extraordinary according to the equation
range of sensitivity of the human ear (a range of over 1,000
million millions to 1), it is helpful to use a logarithmic scale, A = A0e-0.000124t
rather than an absolute scale, to measure sound intensity over where t is time in years and A0 is the amount of 14C present at
this range. The unit of measure is called the decibel, after time t = 0. (See Example 3 in Section 1.5.) Estimate the age
the inventor of the telephone, Alexander Graham Bell. If we of a skull uncovered in an archaeological site if 10% of the
let N be the number of decibels, I the power of the sound in original amount of 14C is still present. [Hint: Find t such that
question (in watts per square centimeter), and I0 the power A = 0.1A0.]
Answers to Matched Problems 5. x = 2 6. x = 5

1
1. (A) 9 = 32 (B) 2 = 41>2 (C) 9 = 3-2 7. (A) - 1.868 734 (B) 3.356 663 (C) Not defined
2. (A) log7 49 = 2 (B) log9 3 = 1
2 (C) log3 1 13 2 = - 1 8. (A) 0.0063 (B) 120.8092
3 1
3. (A) y = 2 (B) x = 3 (C) b = 10 9. (A) 0.8451 (B) 1.7918 (C) 1.1610
2
4. (A) logb R - logb S - logb T (B) 3 1logb R - logb S2 10. 9 yr 11. 70.3%
(C) bu (D) logb x
Chapter 1 Summary and Review

Important Terms, Symbols, and Concepts
1.1 Functions EXAMPLES
• Point-by-point plotting may be used to sketch the graph of an equation in two variables: Plot enough Ex. 1, p. 20
points from its solution set in a rectangular coordinate system so that the total graph is apparent and then
connect these points with a smooth curve.
• A function is a correspondence between two sets of elements such that to each element in the first set
there corresponds one and only one element in the second set. The first set is called the domain and the
set of corresponding elements in the second set is called the range.
• If x is a placeholder for the elements in the domain of a function, then x is called the independent
variable or the input. If y is a placeholder for the elements in the range, then y is called the dependent
variable or the output.
• If in an equation in two variables we get exactly one output for each input, then the equation specifies a Ex. 2, p. 23
function. The graph of such a function is just the graph of the specifying equation. If we get more than
one output for a given input, then the equation does not specify a function.
• The vertical-line test can be used to determine whether or not an equation in two variables specifies a
function (Theorem 1, p. 24).
• The functions specified by equations of the form y = mx + b, where m ∙ 0, are called linear
functions. Functions specified by equations of the form y = b are called constant functions.
• If a function is specified by an equation and the domain is not indicated, we agree to assume that the Ex. 3, p. 25
domain is the set of all inputs that produce outputs that are real numbers. Ex. 5, p. 27
• The symbol f1x2 represents the element in the range of f that corresponds to the element x of the Ex. 4, p. 26
domain. Ex. 6, p. 27
• Break-even and profit–loss analysis use a cost function C and a revenue function R to determine Ex. 7, p. 29
when a company will have a loss 1R 6 C2, will break even 1R = C2, or will have a profit 1R 7 C2.
Typical cost, revenue, profit, and price–demand functions are given on page 28.
1.2 Elementary Functions: Graphs and Transformations
• The graphs of six basic elementary functions (the identity function, the square and cube functions, the Ex. 1, p. 35
square root and cube root functions, and the absolute value function) are shown on page 36.
• Performing an operation on a function produces a transformation of the graph of the function. The Ex. 2, p. 37
basic graph transformations, vertical and horizontal translations (shifts), reflection in the x axis, and Ex. 3, p. 38
vertical stretches and shrinks, are summarized on page 40. Ex. 4, p. 39
• A piecewise-defined function is a function whose definition involves more than one rule. Ex. 5, p. 40
Ex. 6, p. 41
Ex. 7, p. 42
1.3 Quadratic Functions
• If a, b, and c are real numbers with a ∙ 0, then the function Ex. 1, p. 48
f1x2 = ax2 + bx + c Standard form

is a quadratic function in standard form and its graph is a parabola.
Summary and Review 97
• The quadratic formula
- b { 2b2 - 4ac
x = b2 - 4ac Ú 0
2a
can be used to find the x intercepts of a quadratic function.
• Completing the square in the standard form of a quadratic function produces the vertex form
f1x2 = a1x - h2 2 + k Vertex form
• From the vertex form of a quadratic function, we can read off the vertex, axis of symmetry, maximum or Ex. 2, p. 52
minimum, and range, and can easily sketch the graph (page 52). Ex. 3, p. 54
• If a revenue function R1x2 and a cost function C1x2 intersect at a point 1x0, y0 2, then both this point and Ex. 4, p. 55
its x coordinate x0 are referred to as break-even points.
• Quadratic regression on a graphing calculator produces the function of the form y = ax2 + bx + c Ex. 5, p. 57
that best fits a data set.
1.4 Polynomial and Rational Functions
• A polynomial function is a function that can be written in the form
f1x2 = anxn + an - 1xn - 1 + g + a1x + a0

for n a nonnegative integer called the degree of the polynomial. The coefficients a0, a1, c, an are
real numbers with leading coefficient an ∙ 0. The domain of a polynomial function is the set of all
real numbers. Graphs of representative polynomial functions are shown on page 63 and at the back
of the book.
• The graph of a polynomial function of degree n can intersect the x axis at most n times. An x intercept is
also called a zero or root.
• The graph of a polynomial function has no sharp corners and is continuous; that is; it has no holes or
breaks.
• Polynomial regression produces a polynomial of specified degree that best fits a data set. Ex. 1, p. 64
• A rational function is any function that can be written in the form
n1x2
f1x2 = d1x2 ∙ 0
d1x2
where n(x) and d(x) are polynomials. The domain is the set of all real numbers such that
d1x2 ∙ 0. Graphs of representative rational functions are shown on page 66 and at the back
of the book.
• Unlike polynomial functions, a rational function can have vertical asymptotes [but not more than the Ex. 2, p. 66
degree of the denominator d(x)] and at most one horizontal asymptote.
• A procedure for finding the vertical and horizontal asymptotes of a rational function is given on Ex. 3, p. 68
page 68. Ex. 4, p. 69
• An exponential function is a function of the form
f1x2 = bx
where b ∙ 1 is a positive constant called the base. The domain of f is the set of all real numbers, and
the range is the set of positive real numbers.
• The graph of an exponential function is continuous, passes through 10, 12, and has the x axis as a Ex. 1, p. 75
horizontal asymptote. If b 7 1, then bx increases as x increases; if 0 6 b 6 1, then bx decreases as x
increases (Theorem 1, p. 75).
• Exponential functions obey the familiar laws of exponents and satisfy additional properties (Theorem 2,
p. 76).
• The base that is used most frequently in mathematics is the irrational number e ≈ 2.7183.
1.5 Exponential Functions (Continued)

• Exponential functions can be used to model population growth and radioactive decay. Ex. 2, p. 77
• Exponential regression on a graphing calculator produces the function of the form y = ab that best fits Ex. 3, p. 78
x
a data set. Ex. 4, p. 79
• Exponential functions are used in computations of compound interest and continuous compound Ex. 5, p. 80
interest: Ex. 6, p. 81
r mt
A = P a1 + b Compound interest
m
A = Per t Continuous compound interest
(see Summary on page 81).
1.6 Logarithmic Functions
• A function is said to be one-to-one if each range value corresponds to exactly one domain value.
• The inverse of a one-to-one function f is the function formed by interchanging the independent and
dependent variables of f. That is, 1a, b2 is a point on the graph of f if and only if 1b, a2 is a point
on the graph of the inverse of f. A function that is not one-to-one does not have an inverse.
• The inverse of the exponential function with base b is called the logarithmic function with base b,
denoted y = log b x. The domain of log b x is the set of all positive real numbers (which is the range
of bx), and the range of log b x is the set of all real numbers (which is the domain of bx).
• Because log b x is the inverse of the function bx, Ex. 1, p. 86
Logarithmic form Exponential form Ex. 2, p. 87
Ex. 3, p. 87
y = log b x is equivalent to x = by
• Properties of logarithmic functions can be obtained from corresponding properties of exponential Ex. 4, p. 88
functions (Theorem 1, p. 88). Ex. 5, p. 88
Ex. 6, p. 89
• Logarithms to the base 10 are called common logarithms, often denoted simply by log x. Logarithms Ex. 7, p. 89
to the base e are called natural logarithms, often denoted by ln x. Ex. 8, p. 90
Ex. 9, p. 90
• Logarithms can be used to find an investment’s doubling time—the length of time it takes for the value
Ex. 10, p. 91
of an investment to double.
• Logarithmic regression on a graphing calculator produces the function of the form y = a + b ln x that Ex. 11, p. 92
best fits a data set.
Review Exercises
Work through all the problems in this chapter review and check (A) y (B) y
your answers in the back of the book. Answers to all review prob- 5
5
lems are there along with section numbers in italics to indicate
where each type of problem is discussed. Where weaknesses show
up, review appropriate sections in the text.
x x
25 5
A In Problems 1–3, use point-by-point plotting to sketch the graph 25 5
of each equation.
1. y = 5 - x2 25 25
2 2
2. x = y
3. y 2 = 4x2
4. Indicate whether each graph specifies a function:
Review Exercises 99
(C) y (D) y 19. Complete the square and find the standard form for the qua-
dratic function
5 5
f1x2 = - x2 + 4x
Then write a brief verbal description of the relationship
x x between the graph of f and the graph of y = x2.
25 5 25 5
20. Match each equation with a graph of one of the functions f, g,
m, or n in the figure.
25 25 y
f g
2 5
5. For f1x2 = 2x - 1 and g1x2 = x - 2x, find:
(A) f1 - 22 + g1 - 12 (B) f102 # g142
g122 f132 x
(C) (D) 27 7
f132 g122
6. Write in logarithmic form using base e: u = ey.

25
7. Write in exponential form using base e: ln m = k. m n
x
8. Write in logarithmic form using base e: k = e . (A) y = 1x - 22 - 4 2
(B) y = - 1x + 22 2 + 4
9. Write in logarithmic form using base 10: m = 10 z. (C) y = - 1x - 22 2 + 4 (D) y = 1x + 22 2 - 4
Solve Problems 10–12 for x exactly without using a calculator. 21. Referring to the graph of function f in the figure for Problem 20
10. log4 x = 3 11. logx 49 = 2 and using known properties of quadratic functions, find each of
the following to the nearest integer:
12. log3 27 = x
Solve Problems 13–16 for x to three decimal places.
13. 10x = 130.6 14. ex = 200
15. log x = 2.15 16. In x = - 1.147 In Problems 22–25, each equation specifies a function. Determine
whether the function is linear, quadratic, constant, or none of these.
17. Use the graph of function f in the figure to determine (to the
nearest integer) x or y as indicated. 1 + 5x
22. y = 4 - x + 3x2 23. y =
(A) y = f102 (B) 4 = f1x2 6
(C) y = f132 (D) 3 = f1x2 7 - 4x
24. y = 25. y = 8x + 2110 - 4x2
(E) y = f 1 - 62 (F) - 1 = f1x2 2x
f (x)
Solve Problems 26–33 for x exactly without using a calculator.
26. log1x + 52 = log12x - 32 27. 2 In1x - 12 = In1x2 - 52
5
2
28. 9x - 1 = 31 + x 29. e2x = ex -3
30. 12x2ex = 4xex 31. log1>3 9 = x

x 3
27 7 32. logx 8 = - 3 33. log9 x = 2
Solve Problems 34–41 for x to four decimal places.

25 34. x = 31e1.49 2 35. x = 230110- 0.161 2
36. log x = - 2.0144 37. In x = 0.3618
B 18. Sketch a graph of each of the functions in parts (A)–(D)
x
using the graph of function f in the figure below. 38. 35 = 713 2 39. 0.001 = e0.03x
(A) y = - f1x2 (B) y = f1x2 + 4 40. 8,000 = 4,00011.08x 2 41. 52x - 3 = 7.08
(C) y = f1x - 22 (D) y = - f1x + 32 - 3
f (x) 42. Find the domain of each function:
2x - 5 3x
5 (A) f1x2 = 2 (B) g1x2 =
x - x - 6 25 - x
43. Find the vertex form for f1x2 = 4x2 + 4x - 3 and then find
x the intercepts, the vertex, the maximum or minimum, and the
25 5
range.
44. Let f1x2 = ex - 2 and g1x2 = ln 1x + 12. Find all points
25 of intersection for the graphs of f and g. Round answers to
two decimal places.
In Problems 45 and 46, use point-by-point plotting to sketch the In Problems 62–67, discuss the validity of each statement. If the state-
graph of each function. ment is always true, explain why. If not, give a counterexample.
62. Every polynomial function is a rational function.
50 - 66
45. f1x2 = 2
46. f1x2 = 63. Every rational function is a polynomial function.
x + 1 2 + x2
64. The graph of every rational function has at least one vertical
If f1x2 = 5x + 1, find and simplify each of the following in
asymptote.
Problems 47–50.
47. f1f102 2 48. f1f1 - 12 2 65. The graph of every exponential function has a horizontal
asymptote.
49. f12x - 12 50. f14 - x2
66. The graph of every logarithmic function has a vertical
51. Let f1x2 = 3 - 2x. Find asymptote.
(A) f122 (B) f12 + h2
67. There exists a rational function that has both a vertical and
f12 + h2 - f122 horizontal asymptote.
(C) f12 + h2 - f122 (D) ,h ∙ 0
h 68. Sketch the graph of f for x Ú 0.
52. Let f1x2 = 4 - 3x. Find
(A) f132 (B) f13 + h2 9 + 0.3x if 0 … x … 20
f1x2 = e
5 + 0.2x if x 7 20
f13 + h2 - f132
(C) f13 + h2 - f132 (D) ,h ∙ 0 69. Sketch the graph of g for x Ú 0.
h
0.5x + 5 if 0 … x … 10
53. Explain how the graph of m1x2 = - ∙ x - 6∙ is related to
f1x2 = • 1.2x - 2 if 10 6 x … 30
the graph of y = ∙ x ∙ .
2x - 26 if x 7 30
54. Explain how the graph of g1x2 = 0.6x3 + 5 is related to the
graph of y = x3. 70. Write an equation for the graph shown in the form
y = a1x - h2 2 + k, where a is either - 1 or + 1 and h and k
55. The following graph is the result of applying a sequence of are integers.
transformations to the graph of y = x2. Describe the trans-
formations verbally and write an equation for the graph y
y 5
x
25 7
x
25 5
25
25
71. Given f1x2 = - 0.4x2 + 3.2x + 1.2, find the following alge-
56. The graph of a function f is formed by vertically stretching braically (to one decimal place) without referring to a graph:
the graph of y = 1x by a factor of 2, and shifting it to the (A) Intercepts (B) Vertex
left 3 units and down 1 unit. Find an equation for function f
and graph it for - 5 … x … 5 and - 5 … y … 5.
In Problems 57–59, find the equation of any horizontal 72. Graph f1x2 = - 0.4x2 + 3.2x + 1.2 in a graphing calcula-
asymptote. tor and find the following (to one decimal place) using trace
and appropriate commands:
5x + 4 3x2 + 2x - 1
57. f1x2 = 2
58. f1x2 = (A) Intercepts (B) Vertex
x - 3x + 1 4x2 - 5x + 3
x2 + 4 (C) Maximum or minimum (D) Range
59. f1x2 =
100x + 1 C 73. Noting that p = 3.141 592 654 c and
22=1.414 213 562 c explain why the calculator results
In Problems 60 and 61, find the equations of any vertical asymptotes.
shown here are obvious. Discuss similar connections between
x2 + 100 x2 + 3x
60. f1x2 = 2 61. f1x2 = 2
x - 100 x + 2x
Review Exercises 101
the natural logarithmic function and the exponential function (B) Graph S1x2.
with base e.
Energy Charge (June–September)
$3.00 for the first 20 kWh or less
5.70¢ per kWh for the next 180 kWh
3.46¢ per kWh for the next 800 kWh
2.17¢ per kWh for all over 1,000 kWh
84. Money growth. Provident Bank of Cincinnati, Ohio, offered

a certificate of deposit that paid 1.25% compounded quar-
terly. If a $5,000 CD earns this rate for 5 years, how much
will it be worth?
Solve Problems 74–77 exactly without using a calculator.
74. log x - log 8 = log 4 - log1x + 42 85. Money growth. Capital One Bank of Glen Allen, Virginia,
offered a certificate of deposit that paid 1.05% compounded
75. ln1x + 42 - ln12x - 12 = ln x
daily. If a $5,000 CD earns this rate for 5 years, how much
76. ln1x + 62 - ln 2x = 2 ln 2 will it be worth?
77. log 3 x2 = 2 + log 9x 86. Money growth. How long will it take for money invested at
6.59% compounded monthly to triple?
78. Write ln y = - 7t + ln b in an exponential form free of loga-
rithms. Then solve for y in terms of the remaining variables. 87. Money growth. How long will it take for money invested at
7.39% compounded continuously to double?
79. Explain why 1 cannot be used as a logarithmic base.
88. Break-even analysis. The research department in a com-
80. The following graph is the result of applying a sequence of
3 pany that manufactures AM/FM clock radios established the
transformations to the graph of y = 2 x. Describe the trans- following price-demand, cost, and revenue functions:
formations verbally, and write an equation for the graph.
y p1x2 = 50 - 1.25x Price–demand function
5
C1x2 = 160 + 10x Cost function
R1x2 = xp1x2
= x150 - 1.25x2 Revenue function
x
27 7 where x is in thousands of units, and C1x2 and R1x2 are
in thousands of dollars. All three functions have domain
1 … x … 40.
25 (A) Graph the cost function and the revenue function simuta-
neously in the same coordinate system.
81. Given G1x2 = 0.4x2 + 1.6x - 6.5, find the following alge- (B) Determine algebraically when R = C. Then, with the
braically (to three decimal places) without the use of a graph: aid of part (A), determine when R 6 C and R 7 C to
(A) Intercepts (B) Vertex the nearest unit.
(C) Maximum or minimum (D) Range (C) Determine algebraically the maximum revenue (to the
2
82. Graph G1x2 = 0.3x + 1.2x - 6.9 in a standard viewing nearest thousand dollars) and the output (to the nearest unit)
window. Then find each of the following (to one decimal that produces the maximum revenue. What is the wholesale
place) using appropriate commands. price of the radio (to the nearest dollar) at this output?
(A) Intercepts (B) Vertex 89. Profit–loss analysis. Use the cost and revenue functions
from Problem 88.
(A) Write a profit function and graph it in a graphing calculator.
(B) Determine graphically when P = 0, P 6 0, and P 7 0
Applications to the nearest unit.
(C) Determine graphically the maximum profit (to the near-
In all problems involving days, a 365-day year is assumed.
est thousand dollars) and the output (to the nearest unit)
83. Electricity rates. The table shows the electricity rates that produces the maximum profit. What is the wholesale
charged by Easton Utilities in the summer months. price of the radio (to the nearest dollar) at this output?
(A) Write a piecewise definition of the monthly charge S1x2 (in [Compare with Problem 88C.]
dollars) for a customer who uses x kWh in a summer month.
90. Construction. A construction company has 840 feet of Table 2 Price–Supply

chain-link fence that is used to enclose storage areas for
x p = S1x21$2
equipment and materials at construction sites. The supervisor
wants to set up two identical rectangular storage areas shar- 985 30
ing a common fence (see the figure). 2,145 75
2,950 110
4,225 155
5,100 190
x
(D) Use the models in parts (A) and (B) to find the equilibri-
um point. Write the equilibrium price to the nearest cent
y y and the equilibrium quantity to the nearest unit.
Assuming that all fencing is used, 92. Crime statistics. According to data published by the FBI, the
(A) Express the total area A1x2 enclosed by both pens as a crime index in the United States has shown a downward trend
function of x. since the early 1990s (Table 3).
(B) From physical considerations, what is the domain of the (A) Find a cubic regression model for the crime index if
function A? x = 0 represents 1987.
(C) Graph function A in a rectangular coordinate system. (B) Use the cubic regression model to predict the crime
index in 2025.
(D) Use the graph to discuss the number and approximate
locations of values of x that would produce storage areas Table 3 Crime Index
with a combined area of 25,000 square feet.
Year Crimes per 100,000 Inhabitants
(E) Approximate graphically (to the nearest foot) the values
1987 5,550
of x that would produce storage areas with a combined
1992 5,660
area of 25,000 square feet.
1997 4,930
(F) Determine algebraically the dimensions of the storage 2002 4,125
areas that have the maximum total combined area. What 2007 3,749
is the maximum area? 2010 3,350
2013 3,099
91. Equilibrium point. A company is planning to introduce a
10-piece set of nonstick cookware. A marketing company es-
tablished price–demand and price–supply tables for selected 93. Medicine. One leukemic cell injected into a healthy mouse
prices (Tables 1 and 2), where x is the number of cookware will divide into 2 cells in about 12 day. At the end of the day
sets people are willing to buy and the company is willing to these 2 cells will divide into 4. This doubling continues until
sell each month at a price of p dollars per set. 1 billion cells are formed; then the animal dies with leukemic
cells in every part of the body.
(A) Find a quadratic regression model for the data in Table 1.
Estimate the demand at a price level of $180. (A) Write an equation that will give the number N of leuke-
mic cells at the end of t days.
(B) Find a linear regression model for the data in Table 2.
Estimate the supply at a price level of $180. (B) When, to the nearest day, will the mouse die?
(C) Does a price level of $180 represent a stable condition, 94. Marine biology. The intensity of light entering water is
or is the price likely to increase or decrease? Explain. reduced according to the exponential equation
I = I 0e -kd
where I is the intensity d feet below the surface, I0 is the
x p = D1x21$2 intensity at the surface, and k is the coefficient of extinction.
985 330 Measurements in the Sargasso Sea have indicated that half of
2,145 225 the surface light reaches a depth of 73.6 feet. Find k (to five
2,950 170
decimal places), and find the depth (to the nearest foot) at
which 1% of the surface light remains.
4,225 105
5,100 50 95. Agriculture. The number of dairy cows on farms in the
United States is shown in Table 4 for selected years since
1950. Let 1940 be year 0.
Table 4 Dairy Cows on Farms in the 97. Medicare. The annual expenditures for Medicare (in billions
United States of dollars) by the U.S. government for selected years since
Year Dairy Cows (thousands) 1980 are shown in Table 5. Let x represent years since 1980.
1950 23,853 (A) Find an exponential regression model 1y = abx 2 for the
1960 19,527 data. Estimate (to the nearest billion) the annual expendi-
1970 12,091 tures in 2025.
1980 10,758 (B) When will the annual expenditures exceed two trillion
1990 10,015 dollars?
2000 9,190
2010 9,117 Table 5 Medicare Expenditures
Year Billion $
(A) Find a logarithmic regression model 1y = a + b ln x2 1980 37
for the data. Estimate (to the nearest thousand) the num- 1985 72
ber of dairy cows in 2023. 1990 111
(B) Explain why it is not a good idea to let 1950 be year 0. 1995 181
2000 197
96. Population growth. The population of some countries has
2005 299
a relative growth rate of 3% (or more) per year. At this rate,
2010 452
how many years (to the nearest tenth of a year) will it take a
2015 546
population to double?
2 Mathematics
of Finance
2.1 Simple Interest
Introduction
2.2 Compound and
How do I choose the right loan for college? Would it be better to take the
Continuous Compound
Interest dealer’s financing or the rebate for my new car? How much should my parents
offer for the new home they want to buy? To make wise decisions in such mat-
2.3 Future Value of an ters, you need a basic understanding of the mathematics of finance.
Annuity; Sinking Funds
In Chapter 2 we study the mathematics of simple and compound interest,
2.4 Present Value of an ordinary annuities, auto loans, and home mortage loans (see Problems 47–48 in
Annuity; Amortization Section 2.4). You will need a calculator with logarithmic and exponential keys.
A graphing calculator would be even better: It can help you visualize the rate at
which an investment grows or the rate at which principal on a loan is amortized.
You may wish to review arithmetic and geometric sequences, discussed in
Appendix C.2, before beginning this chapter.
Finally, to avoid repeating the following reminder many times, we empha-
size it here: Throughout the chapter, interest rates are to be converted to decimal
form before they are used in a formula.
104
SECTION 2.1 Simple Interest 105
2.1 Simple Interest

■■ The Simple Interest Formula The Simple Interest Formula
■■ Simple Interest and Investments Simple interest is used on short-term notes—often of duration less than 1 year. The
concept of simple interest, however, forms the basis of much of the rest of the mate-
rial developed in this chapter, for which time periods may be much longer than a year.
If you deposit a sum of money P in a savings account or if you borrow a sum of
Reminder money P from a lender, then P is referred to as the principal. When money is borrowed—
To write a percentage as a decimal, whether it is a savings institution borrowing from you when you deposit money in your
divide by 100 and delete the % sign. account, or you borrowing from a lender—a fee is charged for the money borrowed. This
To write a decimal as a percentage, fee is rent paid for the use of another’s money, just as rent is paid for the use of another’s
multiply by 100 and add the % sign. house. The fee is called interest. It is usually computed as a percentage (called the inter-
So 12% and 0.12 are equivalent est rate) of the principal over a given period of time. The interest rate, unless otherwise
expressions. stated, is an annual rate. Simple interest is given by the following formula:
DEFINITION Simple Interest
I = Prt (1)
where I = interest
P = principal
r = annual simple interest rate 1written as a decimal2
t = time in years
For example, the interest on a loan of $100 at 12% for 9 months would be
I = Prt Convert 12% to a decimal (0.12)
9
= 1100210.12210.752 and 9 months to years 1 12 = 0.752.
= $9
At the end of 9 months, the borrower would repay the principal ($100) plus the inter-
est ($9), or a total of $109.
In general, if a principal P is borrowed at a rate r, then after t years, the borrower
will owe the lender an amount A that will include the principal P plus the interest I.
Since P is the amount that is borrowed now and A is the amount that must be paid
back in the future, P is often referred to as the present value and A as the future
value. The formula relating A and P follows:
THEOREM 1 Simple Interest
A = P + Prt
= P11 + rt2 (2)
where A = amount, or future value
P = principal, or present value
r = annual simple interest rate1written as a decimal2
t = time in years
Given any three of the four variables A, P, r, and t in (2), we can solve for the
fourth. The following examples illustrate several types of common problems that can
be solved by using formula (2).
106 CHAPTER 2 Mathematics of Finance
EXAMPLE 1 Total Amount Due on a Loan Find the total amount due on a loan of $800 at 9%
simple interest at the end of 4 months.
SOLUTION To find the amount A (future value) due in 4 months, we use formula (2)
4
with P = 800, r = 0.09, and t = 12 = 13 year. Thus,
A = P11 + rt2
= 800 c1 + 0.09 a 13 bd
= 80011.032
= $824
Matched Problem 1 Find the total amount due on a loan of $500 at 12% simple
interest at the end of 30 months.

(A) Your sister has loaned you $1,000 with the understanding that you will repay the
principal plus 4% simple interest when you can. How much would you owe her
if you repaid the loan after 1 year? After 2 years? After 5 years? After 10 years?
(B) How is the interest after 10 years related to the interest after 1 year? After
2 years? After 5 years?
(C) Explain why your answers are consistent with the fact that for simple interest,
the graph of future value as a function of time is a straight line (Fig. 1).
A
1,600
1,400
1,200
Future value ($)
1,000
800 A 5 1,000(1 1 0.04t)
600
400
200
t
0 1 2 3 4 5 6 7 8 9 10
Time (years)
Figure 1
EXAMPLE 2 Present Value of an Investment If you want to earn an annual rate of 10% on
your investments, how much (to the nearest cent) should you pay for a note that will
be worth $5,000 in 9 months?
SOLUTION We again use formula (2), but now we are interested in finding the prin-
9
cipal P (present value), given A = $5,000, r = 0.1, and t = 12 = 0.75 year. Thus,
A = P11 + rt2 Replace A, r, and t with the
given values, and solve for P.
5,000 = P31 + 0.110.7524
5,000 = 11.0752P
P = $4,651.16
Matched Problem 2 Repeat Example 2 with a time period of 6 months.

CONCEPTUAL INSIGHT
If we consider future value A as a function of time t with the present value P and
the annual rate r being fixed, then A = P + Prt is a linear function of t with y
intercept P and slope Pr. For example, if P = 1,000 and r = 0.04 (Fig. 1), then
A = 1,00011 + 0.04t2 = 1,000 + 40t
is a linear function with y intercept 1,000 and slope 40.
Simple Interest and Investments

Because simple interest is used on short-term notes, the time period is often given
in days rather than months or years. How should a time period given in days be con-
verted to years? In this section, we will divide by 360, assuming a 360-day year called
a banker’s year. In other sections, we will assume a 365-day year. The choice will
always be clearly stated. This assumption does not affect calculations in which the
time period is not given in days. If the time period is given in months or quarters, for
example, we would divide by 12 or 4, respectively, to convert the time period to years.
EXAMPLE 3 Interest Rate Earned on a Note T-bills (Treasury bills) are one of the instru-
ments that the U.S. Treasury Department uses to finance the public debt. If you
buy a 180-day T-bill with a maturity value of $10,000 for $9,893.78, what annual
simple interest rate will you earn? (Express answer as a percentage, correct to three
decimal places.)
SOLUTION Again we use formula (2), but this time we are interested in finding r,
given P = $9,893.78, A = $10,000, and t = 180>360 = 0.5 year.
A = P11 + rt2 Replace P, A, and t with the
given values, and solve for r.
10,000 = 9,893.7811 + 0.5r2
10,000 = 9,893.78 + 4,946.89r
106.22 = 4,946.89r
106.22
r = ≈ 0.02147 or 2.147%
4,946.89
Matched Problem 3 Repeat Example 3, assuming that you pay $9,828.74 for
the T-bill.
EXAMPLE 4 Interest Rate Earned on an Investment Suppose that after buying a new car you
decide to sell your old car to a friend. You accept a 270-day note for $3,500 at 10%
simple interest as payment. (Both principal and interest are paid at the end of 270
days.) Sixty days later you find that you need the money and sell the note to a third
party for $3,550. What annual interest rate will the third party receive for the invest-
ment? Express the answer as a percentage, correct to three decimal places.
SOLUTION
Step 1 Find the amount that will be paid at the end of 270 days to the holder of
the note.
A = P11 + rt2
= $3,500 3 1 + 10.12 1 270
360 2 4
= $3,762.50
Step 2 For the third party, we are to find the annual rate of interest r required to make
$3,550 grow to $3,762.50 in 210 days 1270 - 602; that is, we are to find r
(which is to be converted to 100r%), given A = $3,762.50, P = $3,550, and
t = 210
360.
A = P + Prt Solve for r.

A - P
r =
Pt
3,762.50 - 3,550
r = ≈ 0.102 62 or 10.262%
210
13,5502 a 360 b
Matched Problem 4 Repeat Example 4 assuming that 90 days after it was

initially signed, the note was sold to a third party for $3,500.
Some online discount brokerage firms offer flat rates for trading stock, but many
still charge commissions based on the transaction amount (principal). Table 1 shows
the commission schedule for one of these firms.
Table 1 Commission Schedule
Principal Commission
$0- $2,499 $29 + 1.6% of principal
$2,500- $9,999 $49 + 0.8% of principal
$10,000+ $99 + 0.3% of principal
EXAMPLE 5 Interest on an Investment An investor purchases 50 shares of a stock at $47.52

per share. After 200 days, the investor sells the stock for $52.19 per share. Using
Table 1, find the annual rate of interest earned by this investment. Express the an-
swer as a percentage, correct to three decimal places.
SOLUTION The principal referred to in Table 1 is the value of the stock. The total
cost for the investor is the cost of the stock plus the commission:
47.521502 = $2,376 Principal
29 + 0.01612,3762 = $67.02 Commission, using line 1 of Table 1
2,376 + 67.02 = $2,443.02 Total investment
When the stock is sold, the commission is subtracted from the proceeds of the sale
and the remainder is returned to the investor:
52.191502 = $2,609.50 Principal
49 + 0.00812,609.502 = $69.88 Commission, using line 2 of Table 1
2,609.50 - 69.88 = $2,539.62 Total return
200 5
Now using formula (2) with A = 2,539.62, P = 2,443.02, and t = = , we have
360 9
A = P11 + rt2
5
2,539.62 = 2,443.02 a1 + rb
9
= 2,443.02 + 1,357.23r
96.60 = 1,357.23r
96.60
r = ≈ 0.07117 or 7.117%
1,357.23
Matched Problem 5 Repeat Example 5 if 500 shares of stock were purchased
for $17.64 per share and sold 270 days later for $22.36 per share.
CONCEPTUAL INSIGHT
The commission schedule in Table 1 specifies a piecewise-defined function C
with independent variable p, the principal (see Section 1.2).
29 + 0.016p if 0 … p 6 2,500
C = • 49 + 0.008p if 2,500 … p 6 10,000
99 + 0.003p if 10,000 … p
Two credit card accounts may differ in a number of ways, including annual inter-
est rates, credit limits, minimum payments, annual fees, billing cycles, and even the
methods for calculating interest. A common method for calculating the interest owed
on a credit card is the average daily balance method. In this method, a balance is cal-
culated at the end of each day, incorporating any purchases, credits, or payments that
were made that day. Interest is calculated at the end of the billing cycle on the average
of those daily balances. The average daily balance method is considered in Example 6.
EXAMPLE 6 Credit Card Accounts A credit card has an annual interest rate of 21.99%, and
interest is calculated by the average daily balance method. In a 30-day billing cycle,
purchases of $56.75, $184.36, and $49.19 were made on days 12, 19, and 24, re-
spectively, and a payment of $100.00 was credited to the account on day 10. If the
unpaid balance at the start of the billing cycle was $842.67, how much interest will
be charged at the end of the billing cycle? What will the unpaid balance be at the
start of the next billing cycle?
SOLUTION First calculate the unpaid balance on each day of the billing cycle:
Days 1 - 9: $842.67
Days 10 - 11: $842.67 - $100.00 = $742.67
Days 12 - 18: $742.67 + $56.75 = $799.42
Days 19 - 23: $799.42 + $184.36 = $983.78
Days 24 - 30: $983.78 + $49.19 = $1,032.97
So the unpaid balance was $842.67 for the first 9 days of the billing cycle, $742.67
for the next 2 days, $799.42 for the next 7 days, and so on. To calculate the average
daily balance, we find the sum of the 30 daily balances, and then divide by 30:
Sum: 91$842.672 + 21$742.672 + 71$799.422 + 51$983.782
+ 71$1032.972 = $26,815.00
Average daily balance: $26,815.00>30 = $893.83
To calculate the interest, use the formula I = Prt with P = $893.83, r = 0.2199,
and t = 30>360:
I = Prt = $893.8310.21992130>3602 = $16.38
Therefore, the interest charged at the end of the billing cycle is $16.38, and the
unpaid balance at the start of the next cycle is $1,032.97 + $16.38 = $1,049.35.
Matched Problem 6 A credit card has an annual interest rate of 16.99%, and
interest is calculated by the average daily balance method. In a 30-day billing cycle,
purchases of $345.86 and $246.71 were made on days 9 and 16, respectively, and a
payment of $500.00 was credited to the account on day 15. If the unpaid balance at the
start of the billing cycle was $1,792.19, how much interest will be charged at the end of
the billing cycle? What will the unpaid balance be at the start of the next billing cycle?
Exercises 2.1
W Skills Warm-up Exercises B In Problems 33–40, use formula (2) for the amount to find each of
the indicated quantities.
In Problems 1–4, if necessary, review Section B.1.
33. P = $4,500; r = 10%; t = 1 quarter; A = ?
1. If your state sales tax rate is 5.65%, how much tax will you
pay on a bicycle that sells for $449.99? 34. P = $8,000; r = 6.5%; t = 90 days; A = ?
2. If your state sales tax rate is 8.25%, what is the total cost of a 35. A = $910; r = 16%; t = 13 weeks; P = ?
motor scooter that sells for $1,349.95? 36. A = $6,608; r = 24%; t = 3 quarters; P = ?
3. A baseball team had a 103–58 win–loss record. Find its win- 37. A = $24,780; P = $21,000; t = 8 months; r = ?
ning percentage to the nearest percentage point.
38. A = $22,135; P = $19,000; t = 39 weeks; r = ?
4. A basketball team played 21 games with a winning percent-
age of 81%. How many games did it lose? 39. A = $736; P = $640; r = 15%; t = ?
40. A = $610; P = $600; r = 5%; t = ?
In Problems 5–8, give the slope and y intercept of each line.
(If necessary, review Section A.2.) C In Problems 41–46, solve each formula for the indicated variable.
5. y = 12,000 + 120x 41. I = Prt; for r 42. I = Prt; for P
6. y = 15,000 + 300x 43. A = P + Prt; for P 44. A = P + Prt; for r
7. y = 2,00011 + 0.025x2 45. A = P11 + rt2; for t 46. I = Prt; for t
8. y = 5,00011 + 0.035x2 47. Discuss the similarities and differences in the graphs of future
value A as a function of time t if $1,000 is invested at simple in-
A In Problems 9–16, convert the given interest rate to decimal form
terest at rates of 4%, 8%, and 12%, respectively (see the figure).
if it is given as a percentage, and to a percentage if it is given in
decimal form. A
9. 6.2% 10. 0.085 2,000

A 5 1,000(1 1 0.12t)
1,800 A 5 1,000(1 1 0.08t)
11. 0.137 12. 4.35% 1,600
Future value ($)
1,400 A 5 1,000(1 1 0.04t)

13. 0.25% 14. 0.0019 1,200
1,000
15. 0.0084 16. 0.765% 800
600
In Problems 17–24, convert the given time period to years, in 400
reduced fraction form, assuming a 360-day year [this assumption 200
t
does not affect the number of quarters (4), months (12), or weeks 0 1 2 3 4 5 6 7 8
(52) in a year]. Time (years)
17. 180 days 18. 9 months 48. Discuss the similarities and differences in the graphs of future
19. 4 months 20. 90 days value A as a function of time t for loans of $400, $800, and
$1,200, respectively, each at 7.5% simple interest (see the
21. 5 quarters 22. 6 weeks figure).
23. 24 weeks 24. 7 quarters
A
In Problems 25–32, use formula (1) for simple interest to find 2,000 A 5 1,200(1 1 0.075t)
each of the indicated quantities. 1,800
1,600
Future value ($)
25. P = $300; r = 7%; t = 2 years; I = ? 1,400 A 5 800(1 1 0.075t)

1,200
26. P = $950; r = 9%; t = 1 year; I = ? 1,000
800
27. I = $36; r = 4%; t = 6 months; P = ? 600 A 5 400(1 1 0.075t)
400
28. I = $15; r = 8%; t = 3 quarters; P = ? 200
t
29. I = $48; P = $600; t = 240 days; r = ? 0 1 2 3 4 5 6 7 8
Time (years)
30. I = $28; P = $700; t = 13 weeks; r = ?
31. I = $60; P = $2,400; r = 5%; t = ?
32. I = $84; P = $7,200; r = 3.5%; t = ?
In Problems 65 and 66, assume that the minimum payment on a

Applications* credit card is the greater of $20 or 2% of the unpaid balance.
65. Find the minimum payment on an unpaid balance of
In all problems involving days, a 360-day year is assumed. When $1,215.45.
annual rates are requested as an answer, express the rate as a
percentage, correct to three decimal places, unless directed other- 66. Find the minimum payment on an unpaid balance of $936.24.
wise. Round dollar amounts to the nearest cent. In Problems 67 and 68, assume that the minimum payment on a
49. If $3,000 is loaned for 4 months at a 4.5% annual rate, how credit card is the greater of $27 or 3% of the unpaid balance.
much interest is earned? 67. Find the minimum payment on an unpaid balance of $815.69.
50. If $5,000 is loaned for 9 months at a 6.2% annual rate, how 68. Find the minimum payment on an unpaid balance of $927.38.
much interest is earned?
69. For services rendered, an attorney accepts a 90-day note for
51. How much interest will you have to pay for a 60-day loan of $5,500 at 8% simple interest from a client. (Both interest and
$500, if a 36% annual rate is charged? principal are repaid at the end of 90 days.) Wishing to use her
52. If a 50% annual rate is charged, how much interest will be money sooner, the attorney sells the note to a third party for
owed on a loan of $1,000 for 30 days? $5,560 after 30 days. What annual interest rate will the third
party receive for the investment?
53. A loan of $7,260 was repaid at the end of 8 months. What
size repayment check (principal and interest) was written, if 70. To complete the sale of a house, the seller accepts a 180-day
an 8% annual rate of interest was charged? note for $10,000 at 7% simple interest. (Both interest and
principal are repaid at the end of 180 days.) Wishing to use
54. A loan of $10,000 was repaid at the end of 6 months. What
the money sooner for the purchase of another house, the
amount (principal and interest) was repaid, if a 6.5% annual
seller sells the note to a third party for $10,124 after 60 days.
rate of interest was charged?
What annual interest rate will the third party receive for the
55. A loan of $4,000 was repaid at the end of 10 months with a investment?
check for $4,270. What annual rate of interest was charged?
Use the commission schedule from Company A shown in Table 2
56. A check for $3,097.50 was used to retire a 5-month $3,000 to find the annual rate of interest earned by each investment in
loan. What annual rate of interest was charged? Problems 71 and 72.
57. If you paid $30 to a loan company for the use of $1,000 for
Table 2 Company A
60 days, what annual rate of interest did they charge?
Principal Commission
58. If you paid $120 to a loan company for the use of $2,000 for
Under $3,000 $25 + 1.8% of principal
90 days, what annual rate of interest did they charge?
$3,000- $10,000 $37 + 1.4% of principal
59. A radio commercial for a loan company states: “You only pay Over $10,000 $107 + 0.7% of principal
29¢ a day for each $500 borrowed.” If you borrow $1,500 for
120 days, what amount will you repay, and what annual inter- 71. An investor purchases 200 shares at $14.20 a share, holds the
est rate is the company charging? stock for 39 weeks, and then sells the stock for $15.75 a share.
60. George finds a company that charges 59¢ per day for each 72. An investor purchases 450 shares at $21.40 a share, holds
$1,000 borrowed. If he borrows $3,000 for 60 days, what the stock for 26 weeks, and then sells the stock for $24.60 a
amount will he repay, and what annual interest rate will he pay share.
the company?
Use the commission schedule from Company B shown in Table 3
61. What annual interest rate is earned by a 13-week T-bill with a to find the annual rate of interest earned by each investment in
maturity value of $1,000 that sells for $989.37? Problems 73 and 74.
62. What annual interest rate is earned by a 33-day T-bill with a
maturity value of $1,000 that sells for $996.16? Table 3 Company B
63. What is the purchase price of a 50-day T-bill with a ma- Principal Commission
turity value of $1,000 that earns an annual interest rate of Under $3,000 $32 + 1.8% of principal
5.53%? $3,000- $10,000 $56 + 1% of principal
Over $10,000 $106 + 0.5% of principal
64. What is the purchase price of a 26-week T-bill with a ma-
turity value of $1,000 that earns an annual interest rate of
73. An investor purchases 215 shares at $45.75 a share, holds the
4.903%?
stock for 300 days, and then sells the stock for $51.90 a share.
74. An investor purchases 75 shares at $37.90 a share, holds
*The authors wish to thank Professor Roy Luke of Pierce College and Pro-
fessor Dennis Pence of Western Michigan University for their many useful the stock for 150 days, and then sells the stock for $41.20
suggestions of applications for this chapter. a share.
Many tax preparation firms offer their clients a refund anticipation In Problems 83–86, assume that the annual interest rate on a
loan (RAL). For a fee, the firm will give a client his refund when the credit card is 19.99% and interest is calculated by the average
return is filed. The loan is repaid when the IRS refund is sent to the daily balance method.
firm. The RAL fee is equivalent to the interest charge for a loan. The
83. The unpaid balance at the start of a 30-day billing cycle was
schedule in Table 4 is from a major RAL lender. Use this schedule
$654.71. No purchases were made during the billing cycle
to find the annual rate of interest for the RALs in Problems 75–78.
and a payment of $654.71 was credited to the account on day
21. Find the unpaid balance at the end of the billing cycle.
Table 4
RAL Amount RAL Fee 84. The unpaid balance at the start of a 30-day billing cycle was
$0- $500 $29.00 $1,583.44. No purchases were made during the billing cycle
$501- $1,000 $39.00 and a payment of $1,583.44 was credited to the account on day
21. Find the unpaid balance at the end of the billing cycle.
$1,001- $1,500 $49.00
$1,501- $2,000 $69.00 85. The unpaid balance at the start of a 30-day billing cycle was
$2,001- $5,000 $89.00 $725.38. A purchase of $49.82 was made on day 15. No
payment was made during the billing cycle and a late fee of
75. A client receives a $475 RAL, which is paid back in 20 days. $37 was charged to the account on day 25. Find the unpaid
balance at the end of the billing cycle.
76. A client receives a $1,100 RAL, which is paid back in 30 days.
77. A client receives a $1,900 RAL, which is paid back in 15 days. $475.17. A purchase of $125.93 was made on day 3. No
78. A client receives a $3,000 RAL, which is paid back in 25 days. payment was made during the billing cycle and a late fee of
$37 was charged to the account on day 25. Find the unpaid
In Problems 79–82, assume that the annual interest rate on a balance at the end of the billing cycle.
credit card is 25.74% and interest is calculated by the average
daily balance method. A payday loan or short-term loan is a loan that is repaid either on
the next payday or over a short term. In Problems 87–90, assume
79. The unpaid balance at the start of a 28-day billing cycle was a 365-day year to express the annual interest rate as a percentage,
$955.13. A $5,000 purchase was made on the first day of the rounded to the nearest integer.
billing cycle and a $50 payment was credited to the account
on day 21. How much interest will be charged at the end of 87. In the UK, if you use a short-term loan from Wonga Group
the billing cycle? Limited to borrow £400 for a period of two weeks, you have to
repay £444.80. Find the annual interest rate for this loan.
$955.13. A $50 payment was credited to the account on day 88. In South Africa, if you use a short-term loan from Wonga
21 of the billing cycle and a $5,000 purchase was made on Group Limited to borrow R3,000 for a period of 12 days,
the last day of the billing cycle. How much interest will be you have to repay R3,530.04. Find the annual interest rate
charged at the end of the billing cycle? for this loan.
81. The unpaid balance at the start of a 28-day billing cycle was 89. In Spain, if you use a short-term loan from Wonga Group
$1,472.35. Purchases of $154.15 and $38.76 were made Limited to borrow €250 for a period of 18 days, you have to
on days 5 and 12, respectively, and a payment of $250 was repay €257.43. Find the annual interest rate for this loan.
credited to the account on day 18. Find the unpaid balance at 90. In Spain, if you use a short-term loan from Wonga Group Lim-
the end of the billing cycle. ited to borrow €300 for a period of 25 days, you have to repay
82. The unpaid balance at the start of a 28-day billing cycle was €329.70. Find the annual interest rate for this loan.
$1,837.23. Purchases of $126.54 and $52.89 were made on
days 21 and 27, respectively, and a payment of $100 was Answers to Matched Problems
credited to the account on day 20. Find the unpaid balance at
the end of the billing cycle. 1. $650 2. $4,761.90 3. 3.485%
4. 15.0% 5. 31.439% 6. $26.94; $1,911.70
2.2 Compound and Continuous Compound Interest

■■ Compound Interest Compound Interest
■■ Continuous Compound Interest If at the end of a payment period the interest due is reinvested at the same rate, then
■■ Growth and Time the interest as well as the original principal will earn interest during the next payment
period. Interest paid on interest reinvested is called compound interest.
■■ Annual Percentage Yield
For example, suppose you deposit $1,000 in a bank that pays 8% compounded
quarterly. How much will the bank owe you at the end of a year? Compounding
SECTION 2.2 Compound and Continuous Compound Interest 113
quarterly means that earned interest is paid to your account at the end of each
3-month period and that interest as well as the principal earns interest for the next
quarter. Using the simple interest formula (2) from the preceding section, we com-
pute the amount in the account at the end of the first quarter after interest has been
paid:
A = P11 + rt2
= 1,00031 + 0.081 14 24
= 1,00011.022 = $1,020
Now, $1,020 is your new principal for the second quarter. At the end of the sec-
ond quarter, after interest is paid, the account will have
A = $1,02031 + 0.081 14 24
= $1,02011.022 = $1,040.40
Similarly, at the end of the third quarter, you will have
A = $1,040.4031 + 0.081 14 24
= $1,040.4011.022 = $1,061.21
Finally, at the end of the fourth quarter, the account will have
A = $1,061.2131 + 0.081 14 24
= $1,061.2111.022 = $1,082.43
How does this compounded amount compare with simple interest? The amount
with simple interest would be
A = P11 + rt2
= $1,00031 + 0.081124
= $1,00011.082 = $1,080
We see that compounding quarterly yields $2.43 more than simple interest would
provide.
Let’s look over the calculations for compound interest above to see if we can
uncover a pattern that might lead to a general formula for computing compound
interest:
A = 1,00011.022 End of first quarter
A = 31,00011.022411.022 = 1,00011.022 2 End of second quarter
2 3
A = 31,00011.022 411.022 = 1,00011.022 End of third quarter
A = 31,00011.022 3 411.022 = 1,00011.022 4 End of fourth quarter
It appears that at the end of n quarters, we would have
A = 1,00011.022 n End of nth quarter
or
A = 1,00031 + 0.081 14 24 n
0.08 n
= 1,00031 + 4 4
where 0.08
4 = 0.02 is the interest rate per quarter. Since interest rates are generally
quoted as annual nominal rates, the rate per compounding period is found by
dividing the annual nominal rate by the number of compounding periods per year.
In general, if P is the principal earning interest compounded m times a year at an

annual rate of r, then (by repeated use of the simple interest formula, using i = r>m,
the rate per period) the amount A at the end of each period is
A = P11 + i2 End of first period
A = 3P11 + i2411 + i2 = P11 + i2 2 End of second period
A = 3P11 + i2 2 411 + i2 = P11 + i2 3 End of third period
f
A = 3P11 + i2 n-1 411 + i2 = P11 + i2 n End of nth period
We summarize this important result in Theorem 1:
THEOREM 1 Compound Interest
A = P11 + i2 n (1)
where i = r>m and A = amount 1future value2 at the end of n periods
P = principal 1present value2
r = annual nominal rate*
i = rate per compounding period
n = total number of compounding periods
*
This is often shortened to “annual rate” or just “rate.”
CONCEPTUAL INSIGHT
Formula (1) of Theorem 1 is equivalent to the formula
r mt
A = Pa1 + b (2)
m
where t is the time, in years, that the principal is invested. For a compound inter-
est calculation, formula (2) may seem more natural to use than (1), if r (the annual
interest rate) and t (time in years) are given. On the other hand, if i (the interest
rate per period), and n (the number of compounding periods) are given, formula
(1) may seem easier to use. It is not necessary to memorize both formulas, but it is
important to understand how they are related.
EXAMPLE 1 Comparing Interest for Various Compounding Periods If $1,000 is invested at

8% compounded
(A) annually, (B) semiannually,
(C) quarterly, (D) monthly,
what is the amount after 5 years? Write answers to the nearest cent.
SOLUTION
(A) Compounding annually means that there is one interest payment period per
year. So, n = 5 and i = r = 0.08.
A = P11 + i2 n
= 1,00011 + 0.082 5 Use a calculator.
= 1,00011.469 3282
= $1,469.33 Interest earned = A - p = $469.33.
(B) Compounding semiannually means that there are two interest payment periods
per year. The number of payment periods in 5 years is n = 2152 = 10, and the
interest rate per period is
r 0.08
i = = = 0.04
m 2
A = P11 + i2 n
= 1,00011 + 0.042 10 Use a calculator.
= 1,00011.480 2442
= $1,480.24 Interest earned = A - P = $480.24.
(C) Compounding quarterly means that there are four interest payments per year.
So, n = 4152 = 20 and i = 0.08
4 = 0.02.
A = P11 + i2 n
= 1,00011 + 0.022 20 Use a calculator.
= 1,00011.485 9472
Reminder (D) Compounding monthly means that there are twelve interest payments per year.
The bar over the 6 in i = 0.006 666 So, n = 12152 = 60 and i = 0.08
12 = 0.006 666 (see the Reminder).
indicates a repeating decimal expan- A = P11 + i2 n
sion. Rounding i to a small number
of decimal places, such as 0.007 or 0.08 60
= 1,000 a1 + b Use a calculator.
0.0067, can result in round-off errors. 12
To avoid this, use as many decimal = 1,00011.489 8462
places for i as your calculator is
capable of displaying.
Matched Problem 1 Repeat Example 1 with an annual interest rate of 6%
over an 8-year period.
Continuous Compound Interest

In Example 1, we considered an investment of $1,000 at an annual rate of 8%. We
calculated the amount after 5 years for interest compounded annually, semiannually,
quarterly, and monthly. What would happen to the amount if interest were com-
pounded daily, or every minute, or every second?
Although the difference in amounts in Example 1 between compounding semian-
nually and annually is $1,480.24 - $1,469.33 = $10.91, the difference between
compounding monthly and quarterly is only $1,489.85 - $1,485.95 = $3.90. This
suggests that as the number m of compounding periods per year increases without
bound, the amount will approach some limiting value. To see that this is indeed the
case, we rewrite the amount A as follows:
r
A = P11 + i2 n Substitute i = , n = mt.
m
r mt r
= Pa1 + b Multiply the exponent by 1=12.
m r
r 3m>r4rt m 1 r
= Pa1 + b Let x = ; then = .
m r x m
1 xrt
= Pa1 + b Use a law of exponents: axy = 1ax 2 y.
x
1 x rt
= Pc a1 + b d
x
As the number m of compounding periods increases without bound, so does x. So
$491.76
$491.82
$489.85
the expression in square brackets gets close to the irrational number e ≈ 2.7183 (see
$485.95
$490 $480.24 Table 1 in Section 1.5), and the amount approaches the limiting value
A = Pert = 1,000e0.08152 ≈ $1,491.8247
$480
In other words, no matter how often interest is compounded, the amount in the
$469.33
account after 5 years will never equal or exceed $1,491.83. Therefore, the interest
$470
I1 = A - P2 will never equal or exceed $491.83 (Fig. 1).
$460
CONCEPTUAL INSIGHT
One column in Figure 1 is labeled with the symbol ∞, read as “infinity.” This
1 2 4 12 365 `
Number of compounding periods per year symbol does not represent a real number. We use ∞ to denote the process of al-
Figure 1 Interest on $1,000 for 5 years
lowing m, the number of compounding periods per year, to get larger and larger
at 8% with various compounding periods with no upper limit on its size.
The formula we have obtained, A = Pert, is known as the continuous com-

pound interest formula. It is used when interest is compounded continuously, that
is, when the number of compounding periods per year increases without bound.
THEOREM 2 Continuous Compound Interest Formula

If a principal P is invested at an annual rate r (expressed as a decimal) compound-
ed continuously, then the amount A in the account at the end of t years is given by
A = Pert (3)
EXAMPLE 2 Compounding Daily and Continuously What amount will an account have after
2 years if $5,000 is invested at an annual rate of 8%
(A) compounded daily? (B) compounded continuously?
Compute answers to the nearest cent.
SOLUTION
(A) Use the compound interest formula
r mt
A = Pa1 + b
m
with P = 5,000, r = 0.08, m = 365, and t = 2:
0.08 13652122
A = 5,000 a1 + b Use a calculator.
365
= $5,867.45
(B) Use the continuous compound interest formula
A = Pert
with P = 5,000, r = 0.08, and t = 2:
A = 5,000e10.082122 Use a calculator.
= $5,867.55
In Example 2B, do not use the approximation 2.7183 for e; it
! CAUTION is not accurate enough to compute the correct amount to the
nearest cent. Instead, use your calculator’s built-in e. Avoid any rounding off until
the end of the calculation, when you round the amount to the nearest cent.
Matched Problem 2 What amount will an account have after 1.5 years if
$8,000 is invested at an annual rate of 9%
(A) compounded weekly? (B) compounded continuously?
Compute answers to the nearest cent.
CONCEPTUAL INSIGHT
The continuous compound interest formula A = Pert is identical, except for the
names of the variables, to the equation y = cekt that we used to model population
growth in Section 1.5. Like the growth of an investment that earns continuous com-
pound interest, we usually consider the population growth of a country to be continu-
ous: Births and deaths occur all the time, not just at the end of a month or quarter.
Growth and Time

How much should you invest now to have a given amount at a future date? What annual
rate of return have your investments earned? How long will it take your investment to
double in value? The formulas for compound interest and continuous compound inter-
est can be used to answer such questions. If the values of all but one of the variables in
the formula are known, then we can solve for the remaining variable.
EXAMPLE 3 Finding Present Value How much should you invest now at 10% to have $8,000
toward the purchase of a car in 5 years if interest is
(A) compounded quarterly? (B) compounded continuously?
SOLUTION
(A) We are given a future value A = $8,000 for a compound interest invest-
ment, and we need to find the present value P given i = 0.10
4 = 0.025 and
n = 4152 = 20.
A = P11 + i2 n
8,000 = P11 + 0.0252 20
8,000
P = Use a calculator.
11 + 0.0252 20
8,000
=
1.638 616
= $4,882.17
Your initial investment of $4,882.17 will grow to $8,000 in 5 years.
(B) We are given A = $8,000 for an investment at continuous compound interest,
and we need to find the present value P given r = 0.10 and t = 5.
A = Pert
8,000 = Pe0.10152
8,000
P = Use a calculator.
e0.10152
P = $4,852.25
Your initial investment of $4,852.25 will grow to $8,000 in 5 years.
Matched Problem 3 How much should new parents invest at 8% to have

$80,000 toward their child’s college education in 17 years if interest is
(A) compounded semiannually? (B) compounded continuously?
A graphing calculator is a useful tool for studying compound interest. In Figure 2, we

use a spreadsheet to illustrate the growth of the investment in Example 3A both numeri-
cally and graphically. Similar results can be obtained from most graphing calculators.
Figure 2 Growth of $4,882.17 at 10% compounded quarterly for 5 years
Solving the compound interest formula or the continuous compound interest for-
mula for r enables us to determine the rate of growth of an investment.
EXAMPLE 4 Computing Growth Rate Figure 3 shows that a $10,000 investment in a growth-
oriented mutual fund over a 10-year period would have grown to $126,000. What
annual nominal rate would produce the same growth if interest was:
(A) compounded annually? (B) compounded continuously?
Express answers as percentages, rounded to three decimal places.

$140,000
120,000
$126,000
100,000
Mutual Fund
80,000
60,000 $63,000
40,000
20,000 Russell Index
$10,000
0 1 2 3 4 5 6 7 8 9 10
Years
Figure 3 Growth of a $10,000 investment
SOLUTION
(A) 126,000 = 10,00011 + r2 10
12.6 = 11 + r2 10
10
212.6 = 1 + r
10
r = 212.6 - 1 = 0.28836 or 28.836%
(B) 126,000 = 10,000er1102

12.6 = e10r Take ln of both sides.
ln 12.6 = 10r
ln 12.6
r = = 0.25337 or 25.337%
10
Matched Problem 4 The Russell Index tracks the average performance of
various groups of stocks. Figure 3 shows that, on average, a $10,000 investment in
midcap growth funds over a 10-year period would have grown to $63,000. What
annual nominal rate would produce the same growth if interest were
(A) compounded annually? (B) compounded continuously?
Express answers as percentages, rounded to three decimal places.
CONCEPTUAL INSIGHT
We can solve A = P11 + i2 n for n using a property of logarithms:
logb M p = p logb M
Theoretically, any base can be used for the logarithm, but most calculators only
evaluate logarithms with base 10 (denoted log) or base e (denoted ln).
Finally, if we solve the compound interest formula for n (or the continuous com-
pound interest formula for t), we can determine the growth time of an investment—
the time it takes a given principal to grow to a particular value (the shorter the time,
the greater the return on the investment).
Example 5 illustrates three methods for solving for growth time.
EXAMPLE 5 Computing Growth Time How long will it take $10,000 to grow to $12,000 if it
is invested at 9% compounded monthly?
SOLUTION
Method 1. Use logarithms and a calculator:
A = P11 + i2 n
0.09 n
12,000 = 10,000 a1 + b
12
1.2 = 1.0075n
Now, solve for n by taking logarithms of both sides:
ln 1.2 = ln 1.0075n We choose the natural logarithm (base e)
ln 1.2 = n ln 1.0075 and use the property ln M p = p ln M.
14000
ln 1.2
n =
ln 1.0075
≈ 24.40 ≈ 25 months or 2 years and 1 month
Note: 24.40 is rounded up to 25 to guarantee reaching $12,000 since interest is
paid at the end of each month.
0 50
10000
Method 2. Use a graphing calculator: To solve this problem using graphical approx-
y1 5 10,000(1.0075) x imation techniques, we graph both sides of the equation 12,000 = 10,00011.00752 n
y2 5 12,000 and find that the graphs intersect at x = n = 24.40 months (Fig. 4). So the growth
Figure 4 time is 25 months.
Method 3. Most graphing calculators have an approximation process that is referred

to as an equation solver. Figure 5 shows the equation solver on a TI-84 Plus CE.
After entering values for three of the four variables, the solver will approximate
the value of the remaining variable. Once again, we see that the growth time is 25
months (Fig. 5).
Matched Problem 5 How long will it take $10,000 to grow to $25,000 if it is

invested at 8% compounded quarterly?
Figure 5 TI-84 Plus CE equation solver
Annual Percentage Yield

Table 1 lists the rate and compounding period for certificates of deposit (CDs) offered
by four banks. How can we tell which of these CDs has the best return?
Table 1 Certificates of Deposit (CDs)
Bank Rate Compounded
Advanta 4.93% monthly
DeepGreen 4.95% daily
Charter One 4.97% quarterly
Liberty 4.94% continuously

Determine the value after 1 year of a $1,000 CD purchased from each of the banks in
Table 1. Which CD offers the greatest return? Which offers the least return?
If a principal P is invested at an annual rate r compounded m times a year, then
the amount after 1 year is
r m
A = Pa1 + b
m
The simple interest rate that will produce the same amount A in 1 year is called the
annual percentage yield (APY). To find the APY, we proceed as follows:
amount at amount at
° simple interest ¢ = ° compound interest ¢
after 1 year after 1 year
r m
P11 + APY2 = Pa1 + b Divide both sides by P.
m
r m
1 + APY = a1 + b Isolate APY on the left side.
m
r m
APY = a1 + b - 1
m
If interest is compounded continuously, then the amount after 1 year is A = Per.
So to find the annual percentage yield, we solve the equation
P11 + APY2 = Per
for APY, obtaining APY = er - 1. We summarize our results in Theorem 3.
THEOREM 3 Annual Percentage Yield

If a principal is invested at the annual (nominal) rate r compounded m times a
year, then the annual percentage yield is
r m
APY = a1 + b - 1
m
If a principal is invested at the annual (nominal) rate r compounded continuously,
then the annual percentage yield is
APY = er - 1
The annual percentage yield is also referred to as the effective rate or true interest rate.
Compound rates with different compounding periods cannot be compared directly

(see Explore and Discuss 1). But since the annual percentage yield is a simple interest
rate, the annual percentage yields for two different compound rates can be compared.
EXAMPLE 6 Using APY to Compare Investments Find the APYs (expressed as a percentage, cor-
rect to three decimal places) for each of the banks in Table 1 and compare these CDs.
0.0493 12
SOLUTION Advanta: APY = a1 + b - 1 = 0.05043 or 5.043%
12
0.0495 365
DeepGreen: APY = a1 + b - 1 = 0.05074 or 5.074%
365
0.0497 4
Charter One: APY = a1 + b - 1 = 0.05063 or 5.063%
4
Liberty: APY = e0.0494 - 1 = 0.05064 or 5.064%
Comparing these APYs, we conclude that the DeepGreen CD will have the largest
return and the Advanta CD will have the smallest.
Matched Problem 6 Southern Pacific Bank offered a 1-year CD that paid

4.8% compounded daily and Washington Savings Bank offered one that paid
4.85% compounded quarterly. Find the APY (expressed as a percentage, correct to
three decimal places) for each CD. Which has the higher return?
EXAMPLE 7 Computing the Annual Nominal Rate Given the APY A savings and loan wants
to offer a CD with a monthly compounding rate that has an APY of 7.5%. What
annual nominal rate compounded monthly should it use?
Check with a graphing calculator.
r m
SOLUTION APY = a1 + b - 1
m
r 12
0.075 = a1 + b - 1
12
r 12
1.075 = a1 +
b
12
12 r
21.075 = 1 +
12
12 r
21.075 - 1 =
12
12
r = 121 21.075 - 12 Use a calculator.
= 0.072 539 or 7.254%
So an annual nominal rate of 7.254% compounded monthly is equivalent to an APY

of 7.5%.
CHECK We use an equation solver on a graphing calculator to check this result (Fig. 6).
Matched Problem 7 What is the annual nominal rate compounded quarterly

for a bond that has an APY of 5.8%?
Each compound interest problem involves two interest rates.

Figure 6 TI-84 Plus CE equation solver ! CAUTION Referring to Example 5, r = 0.09 or 9% is the annual nomi-
nal compounding rate, and i = r>12 = 0.0075 or 0.75% is the
interest rate per month. Do not confuse these two rates by using r in place of i in the
compound interest formula. If interest is compounded annually, then i = r>1 = r. In
all other cases, r and i are not the same.

(A) Which would be the better way to invest $1,000: at 9% simple interest for 10
years, or at 7% compounded monthly for 10 years?
(B) Explain why the graph of future value as a function of time is a straight line for
simple interest, but for compound interest the graph curves upward (see Fig. 7).
A
Compound interest
2,000
Simple interest
1,800
Future value ($)
1,600
1,400
1,200
t
2 4 6 8 10
Time (years)
Figure 7
CONCEPTUAL INSIGHT
The two curves in Figure 7 intersect at t = 0 and again near t = 7. The t coordi-
nate of each intersection point is a solution of the equation
1,00011 + 0.09t2 = 1,00011 + 0.07>122 12t
Don’t try to use algebra to solve this equation. It can’t be done. But the solutions
are easily approximated on a graphing calculator (Fig. 8).
2,000
0 10
0
Figure 8
Exercises 2.2
Find all dollar amounts to the nearest cent. When an interest rate In Problems 29–36, use the given interest rate i per compounding
is requested as an answer, express the rate as a percentage correct period to find r, the annual rate.
to two decimal places, unless directed otherwise. In all problems
29. 1.73% per half-year
involving days, use a 365-day year.
30. 2.34% per quarter
W Skills Warm-up Exercises
31. 0.53% per month
In Problems 1–8, solve the equation for the unknown quantity.
(If necessary, review sections B.7, 1.5, and 1.6.) 32. 0.012% per day
1. 1,641.6 = P11.22 3
2. 2,652.25 = P11.032 2 33. 2.19% per quarter
3. 12x3 = 58,956 4. 100x4 = 15,006.25 34. 3.69% per half-year
5. 6.75 = 311 + i2 2 6. 13.72 = 511 + i2 3 35. 0.006% per day
7. 14,641 = 10,00011.12 n 8. 2,488.32 = 1,00011.22 n 36. 0.47% per month
B 37. If $100 is invested at 6% compounded
A In Problems 9–12, use compound interest formula (1) to find
(A) annually (B) quarterly (C) monthly
each of the indicated values.
what is the amount after 4 years? How much interest is
9. P = $5,000; i = 0.005; n = 36; A = ? earned?
10. P = $2,800; i = 0.003; n = 24; A = ? 38. If $2,000 is invested at 7% compounded
11. A = $8,000; i = 0.02; n = 32; P = ? (A) annually (B) quarterly (C) monthly
12. A = $15,000; i = 0.01; n = 28; P = ? what is the amount after 5 years? How much interest is
earned?
In Problems 13–20, use the continuous compound interest formula 39. If $5,000 is invested at 5% compounded monthly, what is the
(3) to find each of the indicated values. amount after
13. P = $2,450; r = 8.12%; t = 3 years; A = ? (A) 2 years? (B) 4 years?
14. P = $995; r = 22%; t = 2 years; A = ? 40. If $20,000 is invested at 4% compounded monthly, what is
the amount after
15. A = $3,450; r = 7.65%; t = 9 years; P = ?
(A) 5 years? (B) 8 years?
16. A = $19,000; r = 7.69%; t = 5 years; P = ? 41. If $8,000 is invested at 7% compounded continuously, what
17. A = $88,000; P = $71,153; r = 8.5%; t = ? is the amount after 6 years?
18. A = $32,982; P = $27,200; r = 5.93%; t = ? 42. If $30,000 is invested at 9.75% compounded continuously,
what is the amount after 10 years?
19. A = $15,875; P = $12,100; t = 48 months; r = ?
43. Discuss the similarities and the differences in the graphs of
20. A = $23,600; P = $19,150; t = 60 months; r = ? future value A as a function of time t if $1,000 is invested for
8 years and interest is compounded monthly at annual rates
In Problems 21–28, use the given annual interest rate r and the com-
of 4%, 8%, and 12%, respectively (see the figure).
pounding period to find i, the interest rate per compounding period.
A
21. 6.6% compounded quarterly
22. 3.84% compounded monthly 3,000
2,800
)
A 5 1,000 1 1
0.12
12 )
12t
23. 10.2% compounded monthly 2,600

Future value ($)
2,400
24. 2.94% compounded semiannually 2,200
2,000 )
A 5 1,000 1 1
0.08
12 )
12t
25. 7.3% compounded daily 1,800

1,600
26. 5.44% compounded quarterly
1,400 )
A 5 1,000 1 1
0.04
12 )
12t
27. 4.86% compounded semiannually 1,200

1,000 t
0 1 2 3 4 5 6 7 8
28. 10.95% compounded daily
Time (years)
44. Discuss the similarities and differences in the graphs of 55. How long will it take $4,000 to grow to $9,000 if it is
future value A as a function of time t for loans of $4,000, invested at 7% compounded monthly?
$8,000, and $12,000, respectively, each at 7.5% compounded
56. How long will it take $7,000 to grow to $9,800 if it is
monthly for 8 years (see the figure).
invested at 5.6% compounded quarterly?
A 57. How long will it take $7,000 to grow to $9,800 if it is
invested at 5.6% compounded continuously?
25,000
22,500
)
A 5 12,000 1 1
0.075
12 )12t
58. How long will it take $42,000 to grow to $60,276 if it is

20,000
invested at 4.25% compounded continuously?
Future value ($)
17,500
15,000 )
A 5 8,000 1 1
0.075
12 ) 12t
C In Problems 59 and 60, use compound interest formula (1) to find

12,500
10,000 n to the nearest larger integer value.
7,500 )
A 5 4,000 1 1
0.075
12 )
12t
59. A = 2P; i = 0.06; n = ?
5,000
2,500 60. A = 3P; i = 0.07; n = ?
0 t
0 1 2 3 4 5 6 7 8
61. How long will it take money to double if it is invested at
Time (years)
(A) 10% compounded quarterly?
45. If $1,000 is invested in an account that earns 9.75% com- (B) 12% compounded quarterly?
pounded annually for 6 years, find the interest earned during 62. How long will it take money to double if it is invested at
each year and the amount in the account at the end of each
(A) 8% compounded semiannually?
year. Organize your results in a table.
(B) 7% compounded semiannually?
46. If $2,000 is invested in an account that earns 8.25% com-
pounded annually for 5 years, find the interest earned during 63. How long will it take money to double if it is invested at
each year and the amount in the account at the end of each (A) 3% compounded continuously?
year. Organize your results in a table. (B) 6% compounded continuously?
47. If an investment company pays 6% compounded semiannu- 64. How long will it take money to triple if it is invested at
ally, how much should you deposit now to have $10,000 (A) 3% compounded continuously?
(A) 5 years from now? (B) 10 years from now?
(B) 6% compounded continuously?
48. If an investment company pays 8% compounded quarterly,
how much should you deposit now to have $6,000
(A) 3 years from now? (B) 6 years from now?
49. If an investment earns 9% compounded continuously, how Applications
much should you deposit now to have $25,000
(A) 36 months from now? (B) 9 years from now? 65. A newborn child receives a $20,000 gift toward college from
her grandparents. How much will the $20,000 be worth in 17
50. If an investment earns 12% compounded continuously, how
years if it is invested at 7% compounded quarterly?
much should you deposit now to have $4,800
(A) 48 months from now? (B) 7 years from now? 66. A person with $14,000 is trying to decide whether to pur-
chase a car now, or to invest the money at 6.5% compounded
51. What is the annual percentage yield (APY) for money
semiannually and then buy a more expensive car. How much
invested at an annual rate of
will be available for the purchase of a car at the end of 3
(A) 3.9% compounded monthly? years?
(B) 2.3% compounded quarterly? 67. What will a $210,000 house cost 10 years from now if the
52. What is the annual percentage yield (APY) for money in- inflation rate over that period averages 3% compounded
vested at an annual rate of annually?
(A) 4.32% compounded monthly? 68. If the inflation rate averages 4% per year compounded annu-
(B) 4.31% compounded daily? ally for the next 5 years, what will a car that costs $17,000
now cost 5 years from now?
53. What is the annual percentage yield (APY) for money
invested at an annual rate of 69. Rental costs for office space have been going up at 4.8%
(A) 5.15% compounded continuously? per year compounded annually for the past 5 years. If office
space rent is now $25 per square foot per month, what were
(B) 5.20% compounded semiannually? the rental rates 5 years ago?
54. What is the annual percentage yield (APY) for money in-
70. In a suburb, housing costs have been increasing at 5.2% per
vested at an annual rate of
year compounded annually for the past 8 years. A house
(A) 3.05% compounded quarterly? worth $260,000 now would have had what value 8 years ago?
(B) 2.95% compounded continuously?
71. (A) If an investment of $100 were made in 1776, and if it make these estimates? The rule of 72 states that the annual
earned 3% compounded quarterly, how much would it compound rate of growth r of an investment that doubles
be worth in 2026? in n years can be approximated by r = 72>n. Construct a
table comparing the exact rate of growth and the approxi-
(B) Discuss the effect of compounding interest monthly,
mate rate provided by the rule of 72 for doubling times
daily, and continuously (rather than quarterly) on the
of n = 6, 7, c, 12 years. Round both rates to one
$100 investment.
decimal place.
(C) Use a graphing calculator to graph the growth of the
82. Refer to Problem 81. Show that the exact annual compound
investment of part (A).
rate of growth of an investment that doubles in n years is
72. (A) Starting with formula (1), derive each of the following given by r = 100121/n - 12. Graph this equation and the
formulas: rule of 72 on a graphing calculator for 5 … n … 20.
A A 1>n ln A - ln P Solve Problems 83–86 using graphical approximation techniques

P = n, i = a b - 1, n = on a graphing calculator.
11 + i2 P ln11 + i2
83. How long does it take for a $2,400 investment at 13% com-
(B) Explain why it is unnecessary to memorize the formulas pounded quarterly to be worth more than a $3,000 investment
above for P, i, and n if you know formula (1). at 6% compounded quarterly?
73. A promissory note will pay $50,000 at maturity 6 years from 84. How long does it take for a $4,800 investment at 8% com-
now. If you pay $28,000 for the note now, what rate compounded monthly to be worth more than a $5,000 investment
pounded continuously would you earn? at 5% compounded monthly?
74. If you deposit $10,000 in a savings account now, what rate 85. One investment pays 10% simple interest and another pays
compounded continuously would be required for you to with- 7% compounded annually. Which investment would you
draw $12,500 at the end of 4 years? choose? Why?
75. You have saved $7,000 toward the purchase of a car costing 86. One investment pays 9% simple interest and another pays 6%
$9,000. How long will the $7,000 have to be invested at 9% compounded monthly. Which investment would you choose?
compounded monthly to grow to $9,000? (Round up to the Why?
next-higher month if not exact.)
87. What is the annual nominal rate compounded daily for a
76. A married couple has $15,000 toward the purchase of a bond that has an annual percentage yield of 3.39%?
house. For the house that the couple wants to buy, a down
payment of $20,000 is required. How long will the money 88. What is the annual nominal rate compounded monthly for a
have to be invested at 7% compounded quarterly to grow to bond that has an annual percentage yield of 2.95%?
$20,000? (Round up to the next-higher quarter if not exact.) 89. What annual nominal rate compounded monthly has the same
77. An Individual Retirement Account (IRA) has $20,000 in annual percentage yield as 7% compounded continuously?
it, and the owner decides not to add any more money to the 90. What annual nominal rate compounded continuously has
account other than interest earned at 6% compounded daily. the same annual percentage yield as 6% compounded
How much will be in the account 35 years from now when monthly?
the owner reaches retirement age?
Problems 91–94 refer to zero coupon bonds. A zero coupon bond
78. If $1 had been placed in a bank account in the year 1066 and is a bond that is sold now at a discount and will pay its face value
forgotten until now, how much would be in the account at the at some time in the future when it matures—no interest payments
end of 2026 if the money earned 2% interest compounded are made.
annually? 2% simple interest? (Now you can see the power
of compounding and why inactive accounts are closed after a 91. A zero coupon bond with a face value of $30,000 matures in
relatively short period of time.) 15 years. What should the bond be sold for now if its rate of
return is to be 4.348% compounded annually?
79. How long will it take money to double if it is invested at 7%
compounded daily? 8.2% compounded continuously? 92. A zero coupon bond with a face value of $20,000 matures in
10 years. What should the bond be sold for now if its rate of
80. How long will it take money to triple if it is invested at 5% return is to be 4.194% compounded annually?
compounded daily? 6% compounded continuously?
93. If you pay $4,126 for a 20-year zero coupon bond with a
81. In a conversation with a friend, you note that you have two face value of $10,000, what is your annual compound rate of
real estate investments, one that has doubled in value in the return?
past 9 years and another that has doubled in value in the past
12 years. Your friend says that the first investment has been 94. If you pay $32,000 for a 5-year zero coupon bond with a
growing at approximately 8% compounded annually and the face value of $40,000, what is your annual compound rate of
second at 6% compounded annually. How did your friend return?
The buying and selling commission schedule shown in the table 97. An investor purchases 200 shares of stock at $28 per share, holds
is from an online discount brokerage firm. Taking into consider- the stock for 4 years, and then sells the stock for $55 a share.
ation the buying and selling commissions in this schedule, find the
98. An investor purchases 400 shares of stock at $48 per share,
annual compound rate of interest earned by each investment in
holds the stock for 6 years, and then sells the stock for $147
Problems 95–98.
a share.
Transaction Size Commission Rate Answers to Matched Problems

$0–$1,500 $29 + 2.5% of principal
1. (A) $1,593.85 (B) $1,604.71
$1,501–$6,000 $57 + 0.6% of principal
(C) $1,610.32 (D) $1,614.14
$6,001–$22,000 $75 + 0.30% of principal
2. (A) $9,155.23 (B) $9,156.29
$22,001–$50,000 $97 + 0.20% of principal
3. (A) $21,084.17 (B) $20,532.86
$50,001–$500,000 $147 + 0.10% of principal
4. (A) 20.208% (B) 18.405%
$500,001+ $247 + 0.08% of principal
5. 47 quarters, or 11 years and 3 quarters
95. An investor purchases 100 shares of stock at $65 per share, holds 6. Southern Pacific Bank: 4.917%
the stock for 5 years, and then sells the stock for $125 a share. Washington Savings Bank: 4.939%
96. An investor purchases 300 shares of stock at $95 per share, holds Washington Savings Bank has the higher return.
the stock for 3 years, and then sells the stock for $156 a share. 7. 5.678%
2.3 Future Value of an Annuity; Sinking Funds

■■ Future Value of an Annuity Future Value of an Annuity
■■ Sinking Funds An annuity is any sequence of equal periodic payments. If payments are made at the
■■ Approximating Interest Rates end of each time interval, then the annuity is called an ordinary annuity. We con-
sider only ordinary annuities in this book. The amount, or future value, of an annuity
is the sum of all payments plus all interest earned.
Suppose you decide to deposit $100 every 6 months into an account that pays 6%
compounded semiannually. If you make six deposits, one at the end of each interest
payment period, over 3 years, how much money will be in the account after the last
deposit is made? To solve this problem, let’s look at it in terms of a time line. Using
the compound amount formula A = P11 + i2 n, we can find the value of each de-
posit after it has earned compound interest up through the sixth deposit, as shown in
Figure 1.
Now 1 yr 2 yr 3 yr Years
0 1 2 3 4 5 6 Number of periods
$100 $100 $100 $100 $100 $100 Deposits
$100(1.03)
$100(1.03)2
$100(1.03)3 Future value
$100(1.03)4
$100(1.03)5
Figure 1
We could, of course, evaluate each of the future values in Figure 1 using a calcu-
lator and then add the results to find the amount in the account at the time of the sixth
deposit—a tedious project at best. Instead, we take another approach, which leads
directly to a formula that will produce the same result in a few steps (even when the
number of deposits is very large). We start by writing the total amount in the account
after the sixth deposit in the form
SECTION 2.3 Future Value of an Annuity; Sinking Funds 127
S = 100 + 10011.032 + 10011.032 2 + 10011.032 3 + 10011.032 4 + 10011.032 5 (1)

We would like a simple way to sum these terms. Let us multiply each side of (1) by
1.03 to obtain
1.03S = 10011.032 + 10011.032 2 + 10011.032 3 + 10011.032 4 + 10011.032 5 + 10011.032 6 (2)
Subtracting equation (1) from equation (2), left side from left side and right side from
right side, we obtain
1.03S - S = 10011.032 6 - 100 Factor each side.
0.03S = 100311.032 6 - 14 Divide by 0.03.
6
11 + 0.032 - 1
S = 100 We write S in this form to
0.03 observe a general pattern. (3)
In general, if R is the periodic deposit, i the rate per period, and n the number of
periods, then the future value is given by
S = R + R11 + i2 + R11 + i2 2 + g + R11 + i2 n - 1 Note how this

(4)
compares to (1).
and proceeding as in the above example, we obtain the general formula for the future
value of an ordinary annuity:
11 + i2 n - 1
S = R Note how this compares to (3). (5)
i
Returning to the example above, we use a calculator to complete the problem:
11.032 6 - 1 For improved accuracy, keep all values in the
S = 100 calculator until the end; round to the
0.03
required number of decimal places.
= $646.84
CONCEPTUAL INSIGHT
In general, an expression of the form
a + ar + ar 2 + g + ar n - 1
is called a finite geometric series (each term is obtained from the preceding term
by multiplying by r). The sum of the terms of a finite geometric series is (see
Section C.2)
rn - 1
a + ar + ar 2 + g + ar n-1 = a
r - 1
If a = R and r = 1 + i, then equation (4) is the sum of the terms of a finite geo-
metric series and, using the preceding formula, we have
S = R + R11 + i2 + R11 + i2 2 + g + R11 + i2 n - 1

11 + i2 n - 1
= R a = R, r = 1 + i
1 + i - 1
11 + i2 n - 1
= R (5)
i
So formula (5) is a direct consequence of the sum formula for a finite geometric
series.
It is common to use FV (future value) for S and PMT (payment) for R in for-
mula (5). Making these changes, we have the formula in Theorem 1.
THEOREM 1 Future Value of an Ordinary Annuity

11 + i2 n - 1
FV = PMT (6)
i
where FV = future value 1amount2
PMT = periodic payment
i = rate per period
n = number of payments 1periods2
Note: Payments are made at the end of each period.
EXAMPLE 1 Future Value of an Ordinary Annuity What is the value of an annuity at the end
of 20 years if $2,000 is deposited each year into an account earning 8.5% com-
pounded annually? How much of this value is interest?
SOLUTION To find the value of the annuity, use formula (6) with PMT = $2,000,
i = r = 0.085, and n = 20.
11 + i2 n - 1
FV = PMT
i
11.0852 20 - 1
= 2,000 = $96,754.03 Use a calculator.
0.085
To find the amount of interest earned, subtract the total amount deposited in the
annuity (20 payments of $2,000) from the total value of the annuity after the 20th
payment.
Deposits = 2012,0002 = $40,000
Interest = value - deposits = 96,754.03 - 40,000 = $56,754.03
Figure 2, which was generated using a spreadsheet, illustrates the growth of this
account over 20 years.
Figure 2 Ordinary annuity at 8.5% compounded annually for 20 years

Matched Problem 1 What is the value of an annuity at the end of 10 years

if $1,000 is deposited every 6 months into an account earning 8% compounded
semiannually? How much of this value is interest?
The table in Figure 2 is called a balance sheet. Let’s take a closer look at the
construction of this table. The first line is a special case because the payment is made
at the end of the period and no interest is earned. Each subsequent line of the table is
computed as follows:
payment + interest + old balance = new balance
2,000 + 0.08512,0002 + 2,000 = 4,170 Period 2
2,000 + 0.08514,1702 + 4,170 = 6,524.45 Period 3
And so on. The amounts at the bottom of each column in the balance sheet agree with the
results we obtained by using formula (6), as you would expect. Although balance sheets
are appropriate for certain situations, we will concentrate on applications of formula (6).
There are many important problems that can be solved only by using this formula.

(A) Discuss the similarities and differences in the graphs of future value FV as a
function of time t for ordinary annuities in which $100 is deposited each month
for 8 years and interest is compounded monthly at annual rates of 4%, 8%, and
12%, respectively (Fig. 3).
(B) Discuss the connections between the graph of the equation y = 100t, where t is
time in months, and the graphs of part (A).
A
Future value at 12%
16,000
Future value at 8%
14,000
12,000 Future value at 4%
Future value ($)
10,000
8,000
6,000
4,000
2,000
0 t
0 12 24 36 48 60 72 84 96
Time (months)
Figure 3
Sinking Funds
The formula for the future value of an ordinary annuity has another important applica-
tion. Suppose the parents of a newborn child decide that on each of the child’s birthdays
up to the 17th year, they will deposit $PMT in an account that pays 6% compounded
annually. The money is to be used for college expenses. What should the annual deposit
($PMT) be in order for the amount in the account to be $80,000 after the 17th deposit?
We are given FV, i, and n in formula (6), and we must find PMT:
11 + i2 n - 1
FV = PMT
i
11.062 17 - 1
80,000 = PMT Solve for PMT.
0.06
0.06
PMT = 80,000 Use a calculator.
11.062 17 - 1
= $2,835.58 per year
An annuity of 17 annual deposits of $2,835.58 at 6% compounded annually will

amount to $80,000 in 17 years.
This is an example of a sinking fund problem. In general, any account that is
established for accumulating funds to meet future obligations or debts is called a
sinking fund. If the payments are to be made in the form of an ordinary annuity, then
we have only to solve formula (6) for the sinking fund payment PMT:
i
PMT = FV (7)
11 + i2 n - 1
It is important to understand that formula (7), which is convenient to use, is sim-
ply a variation of formula (6). You can always find the sinking fund payment by
first substituting the appropriate values into formula (6) and then solving for PMT,
as we did in the college fund example discussed above. Or you can substitute
directly into formula (7), as we do in the next example. Use whichever method is
easier for you.
EXAMPLE 2 Computing the Payment for a Sinking Fund A company estimates that it will
have to replace a piece of equipment at a cost of $800,000 in 5 years. To have this
money available in 5 years, a sinking fund is established by making equal monthly
payments into an account paying 6.6% compounded monthly.
(A) How much should each payment be?
(B) How much interest is earned during the last year?
SOLUTION
(A) To find PMT, we can use either formula (6) or (7). We choose formula (7) with
FV = $800,000, i = 0.066 #
12 = 0.0055, and n = 12 5 = 60:
i
PMT = FV
11 + i2 n - 1
0.0055
= 800,000
11.00552 60 - 1
= $11,290.42 per month
(B) To find the interest earned during the fifth year, we first use formula (6) with
PMT = $11,290.42, i = 0.0055, and n = 12 # 4 = 48 to find the amount in
the account after 4 years:
11 + i2 n - 1
FV = PMT
i
11.00552 48 - 1
= 11,290.42
0.0055
= $618,277.04 Amount after 4 years
During the 5th year, the amount in the account grew from $618,277.04 to $800,000.
A portion of this growth was due to the 12 monthly payments of $11,290.42. The
remainder of the growth was interest. Thus,
800,000 - 618,277.04 = 181,722.96 Growth in the 5th year
12 # 11,290.42 = 135,485.04 Payments during the 5th year
181,722.96 - 135,485.04 = $46,237.92 Interest during the 5th year
Matched Problem 2 A bond issue is approved for building a marina in a city.

The city is required to make regular payments every 3 months into a sinking fund
paying 5.4% compounded quarterly. At the end of 10 years, the bond obligation
will be retired with a cost of $5,000,000.
(A) What should each payment be?
(B) How much interest is earned during the 10th year?
EXAMPLE 3 Growth in an IRA Jane deposits $2,000 annually into a Roth IRA that earns
6.85% compounded annually. (The interest earned by a Roth IRA is tax free.) Due
to a change in employment, these deposits stop after 10 years, but the account con-
tinues to earn interest until Jane retires 25 years after the last deposit was made.
How much is in the account when Jane retires?
SOLUTION First, we use the future value formula with PMT = $2,000, i = 0.0685,
and n = 10 to find the amount in the account after 10 years:
11 + i2 n - 1
FV = PMT
i
11.06852 10 - 1
= 2,000
0.0685
= $27,437.89
Now we use the compound interest formula from Section 2.2 with P = $27,437.89,
i = 0.0685, and n = 25 to find the amount in the account when Jane retires:
A = P11 + i2 n
= 27,437.8911.06852 25
= $143,785.10
Matched Problem 3 Refer to Example 3. Mary starts a Roth IRA earning the
same rate of interest at the time Jane stops making payments into her IRA. How
much must Mary deposit each year for the next 25 years in order to have the same
amount at retirement as Jane?

Refer to Example 3 and Matched Problem 3. What was the total amount Jane
deposited in order to have $143,785.10 at retirement? What was the total amount
Mary deposited in order to have the same amount at retirement? Do you think it is
advisable to start saving for retirement as early as possible?
Approximating Interest Rates

Algebra can be used to solve the future value formula (6) for PMT or n but not for i.
However, graphical techniques or equation solvers can be used to approximate i to as
many decimal places as desired.
EXAMPLE 4 Approximating an Interest Rate A person makes monthly deposits of $100 into
an ordinary annuity. After 30 years, the annuity is worth $160,000. What annual rate
compounded monthly has this annuity earned during this 30-year period? Express
the answer as a percentage, correct to two decimal places.
SOLUTION Substituting FV = $160,000, PMT = $100, and n = 301122 = 360

in (6) produces the following equation:
11 + i2 360 - 1
160,000 = 100
i
We can approximate the solution to this equation by using graphical techniques
(Figs. 4A, 4B) or an equation solver (Fig. 4C). From Figure 4B or 4C, we see
that i = 0.006 956 7 and 121i2 = 0.083 480 4. So the annual rate (to two decimal
places) is r = 8.35%.
200000
0 0.01
0
(A) (B) (C)

Figure 4
Matched Problem 4 A person makes annual deposits of $1,000 into an ordi-

nary annuity. After 20 years, the annuity is worth $55,000. What annual compound
rate has this annuity earned during this 20-year period? Express the answer as a
percentage, correct to two decimal places.
Exercises 2.3
W Skills Warm-up Exercises 12. Semiannual deposits of $1,900 are made for 7 years into an
annuity that pays 8.5% compounded semiannually.
In Problems 1–8, find the sum of the finite geometric series
a + ar + ar 2 + g + ar n - 1. (If necessary, review Section C.2.) 13. Annual deposits of $1,750 are made for 15 years into an
annuity that pays 6.25% compounded annually.
1. 1 + 2 + 4 + 8 + g + 29
14. Quarterly deposits of $1,200 are made for 18 years into an
2. 1 + 5 + 25 + 125 + g + 58 annuity that pays 7.6% compounded quarterly.
3. a = 30, r = 1, n = 100 B In Problems 15–22, use the future value formula (6) to find each
4. a = 25, r = - 1, n = 81 of the indicated values.
5. a = 10, r = 3, n = 15 15. n = 20; i = 0.03; PMT = $500; FV = ?
6. a = 5, r = 7, n = 8 16. n = 25; i = 0.04; PMT = $100; FV = ?
17. FV = $5,000; n = 15; i = 0.01; PMT = ?
A In Problems 7–14, find i (the rate per period) and n (the number
of periods) for each annuity. 18. FV = $3,500; n = 16; i = 0.07; PMT = ?
7. Quarterly deposits of $500 are made for 20 years into an 19. FV = $4,000; i = 0.02; PMT = 200; n = ?
annuity that pays 8% compounded quarterly. 20. FV = $7,500; i = 0.06; PMT = 300; n = ?
8. Monthly deposits of $350 are made for 6 years into an 21. FV = $7,600; PMT = $500; n = 10; i = ?
annuity that pays 6% compounded monthly. (Round answer to two decimal places.)
9. Semiannual deposits of $900 are made for 12 years into an 22. FV = $4,100; PMT = $100; n = 20; i = ?
annuity that pays 7.5% compounded semiannually. (Round answer to two decimal places.)
C
10. Annual deposits of $2,500 are made for 15 years into an
annuity that pays 6.25% compounded annually. 23. Explain what is meant by an ordinary annuity.
11. Monthly deposits of $345 are made for 7 years into an 24. Explain why no interest is credited to an ordinary annuity at
annuity that pays 7.8% compounded monthly. the end of the first period.
25. Solve the future value formula (6) for n. 39. Bob makes his first $1,000 deposit into an IRA earning
6.4% compounded annually on his 24th birthday and his last
26. Solve the future value formula (6) for i if n = 2.
$1,000 deposit on his 35th birthday (12 equal deposits in all).
With no additional deposits, the money in the IRA continues
to earn 6.4% interest compounded annually until Bob retires
Applications on his 65th birthday. How much is in the IRA when Bob
retires?
27. Guaranty Income Life offered an annuity that pays 6.65%
40. Refer to Problem 39. John procrastinates and does not
compounded monthly. If $500 is deposited into this annuity
make his first $1,000 deposit into an IRA until he is 36, but
every month, how much is in the account after 10 years? How
then he continues to deposit $1,000 each year until he is 65
much of this is interest?
(30 deposits in all). If John’s IRA also earns 6.4% com-
28. Nationwide Bank in the UK offered an annuity that pays pounded annually, how much is in his IRA when he makes
6.95% compounded monthly. If $800 is deposited into this his last deposit on his 65th birthday?
annuity every month, how much is in the account after
41. Refer to Problems 39 and 40. How much would John have to
12 years? How much of this is interest?
deposit each year in order to have the same amount at retire-
29. In order to accumulate enough money for a down payment ment as Bob has?
on a house, a couple deposits $300 per month into an account
42. Refer to Problems 39 and 40. Suppose that Bob decides to
paying 6% compounded monthly. If payments are made at
continue to make $1,000 deposits into his IRA every year
the end of each period, how much money will be in the
until his 65th birthday. If John still waits until he is 36 to start
account in 5 years?
his IRA, how much must he deposit each year in order to
30. A self-employed person has a Keogh retirement plan. (This have the same amount at age 65 as Bob has?
type of plan is free of taxes until money is withdrawn.) If
43. Bank Australia offered an Online Saver account with an APY
deposits of $7,500 are made each year into an account paying
of 1.663%.
8% compounded annually, how much will be in the account
after 20 years? (A) If interest is compounded monthly, what is the equiva-
lent annual nominal rate?
31. A bank in the Netherlands offered an annuity that pays 6.15%
compounded monthly. What equal monthly deposit should be (B) If you wish to have $40,000 in this account after 6 years,
made into this annuity in order to have $125,000 in 25 years? what equal deposit should you make each month?
32. A bank in Belgium offered an annuity that pays 5.95% 44. RaboDirect, a division of Rabobank Australia Limited, offered
compounded monthly. What equal monthly deposit should be a High Interest Savings Account with an APY of 3.093%.
made into this annuity in order to have $250,000 in 20 years?
(A) If interest is compounded monthly, what is the equiva-
33. A company estimates that it will need $100,000 in 8 years to lent annual nominal rate?
replace a computer. If it establishes a sinking fund by making
(B) If a company wishes to have $250,000 in this account
fixed monthly payments into an account paying 7.5% com-
after 5 years, what equal deposit should be made each
pounded monthly, how much should each payment be?
month?
34. Parents have set up a sinking fund in order to have $120,000
45. You can afford monthly deposits of $200 into an account
in 15 years for their children’s college education. How much
that pays 5.7% compounded monthly. How long will it be until
should be paid semiannually into an account paying 6.8%
you have $7,000? (Round to the next-higher month
compounded semiannually?
if not exact.)
35. If $1,000 is deposited at the end of each year for 5 years into
an ordinary annuity earning 8.32% compounded annually, 46. A company establishes a sinking fund for upgrading office
construct a balance sheet showing the interest earned during equipment with monthly payments of $2,000 into an account
each year and the balance at the end of each year. paying 6.6% compounded monthly. How long will it be
before the account has $100,000? (Round up to the next-
36. If $2,000 is deposited at the end of each quarter for 2 years higher month if not exact.)
into an ordinary annuity earning 7.9% compounded quarterly,
construct a balance sheet showing the interest earned during In Problems 47–50, use graphical approximation techniques or an
each quarter and the balance at the end of each quarter. equation solver to approximate the desired interest rate. Express
each answer as a percentage, correct to two decimal places.
37. Beginning in January, a person plans to deposit $100 at the
end of each month into an account earning 6% compounded 47. A person makes annual payments of $1,000 into an ordinary
monthly. Each year taxes must be paid on the interest earned annuity. At the end of 5 years, the amount in the annuity is
during that year. Find the interest earned during each year for $5,840. What annual nominal compounding rate has this
the first 3 years. annuity earned?
38. If $500 is deposited each quarter into an account paying 8% 48. A person invests $2,000 annually in an IRA. At the end of
compounded quarterly for 3 years, find the interest earned 6 years, the amount in the fund is $14,000. What annual
during each of the 3 years. nominal compounding rate has this fund earned?
49. An employee opens a credit union account and deposits $120 52. When would an ordinary annuity consisting of monthly pay-
at the end of each month. After one year, the account contains ments of $200 at 5% compounded monthly be worth more
$1,444.96. What annual nominal rate compounded monthly than a principal of $10,000 invested at 7.5% compounded
has the account earned? monthly?
50. An employee opens a credit union account and deposits
$90 at the end of each month. After two years, the account Answers to Matched Problems
contains $2,177.48. What annual nominal rate compounded 1. Value: $29,778.08; interest: $9,778.08
monthly has the account earned? 2. (A) $95,094.67 (B) $248,628.88
In Problems 51 and 52, use graphical approximation techniques 3. $2,322.73 4. 9.64%
to answer the questions.
51. When would an ordinary annuity consisting of quarterly pay-
ments of $500 at 6% compounded quarterly be worth more
than a principal of $5,000 invested at 4% simple interest?
2.4 Present Value of an Annuity; Amortization

■■ Present Value of an Annuity Present Value of an Annuity
■■ Amortization How much should you deposit in an account paying 6% compounded semiannually
■■ Amortization Schedules in order to be able to withdraw $1,000 every 6 months for the next 3 years? (After the
last payment is made, no money is to be left in the account.)
■■ General Problem-Solving Strategy
Actually, we are interested in finding the present value of each $1,000 that is
paid out during the 3 years. We can do this by solving for P in the compound interest
formula:
A = P11 + i2 n
A
P = = A11 + i2 -n
11 + i2 n
The rate per period is i = 0.06

2 = 0.03. The present value P of the first payment is
1,00011.032 -1, the present value of the second payment is 1,00011.032 -2, and so on.
Figure 1 shows this in terms of a time line.
Years Now 1 yr 2 yr 3 yr
Number of periods 0 1 2 3 4 5 6
Payments $1,000 $1,000 $1,000 $1,000 $1,000 $1,000
$1,000(1.03)21
$1,000(1.03)22
$1,000(1.03)23
Present value
$1,000(1.03)24
$1,000(1.03)25
$1,000(1.03)26
Figure 1
We could evaluate each of the present values in Figure 1 using a calculator and
add the results to find the total present values of all the payments (which will be the
amount needed now to buy the annuity). Since this is a tedious process, particularly
when the number of payments is large, we will use the same device we used in the
preceding section to produce a formula that will accomplish the same result in a
couple of steps. We start by writing the sum of the present values in the form
SECTION 2.4 Present Value of an Annuity; Amortization 135
P = 1,00011.032 -1 + 1,00011.032 -2 + g + 1,00011.032 -6 (1)

Multiplying both sides of equation (1) by 1.03, we obtain
1.03P = 1,000 + 1,00011.032 -1 + g + 1,00011.032 -5 (2)
Now subtract equation (1) from equation (2):
1.03P - P = 1,000 - 1,00011.032 -6 Factor each side.
0.03P = 1,00031 - 11 + 0.032 -6 4 Divide by 0.03.
1 - 11 + 0.032 -6 We write P in this form

P = 1,000 to observe a general pattern. (3)
0.03
In general, if R is the periodic payment, i the rate per period, and n the number of
periods, then the present value of all payments is given by
Note how this
P = R11 + i2 -1 + R11 + i2 -2 + g + R11 + i2 -n
compares to (1).
Proceeding as in the above example, we obtain the general formula for the present
value of an ordinary annuity:
1 - 11 + i2 -n
P = R Note how this compares to (3). (4)
i
Returning to the preceding example, we use a calculator to complete the problem:
1 - 11.032 -6
P = 1,000
0.03
= $5,417.19
CONCEPTUAL INSIGHT
Formulas (3) and (4) can also be established by using the sum formula for a finite
geometric series (see Section C.2):
rn - 1
a + ar + ar 2 + g + ar n - 1 = a
r - 1
It is common to use PV (present value) for P and PMT (payment) for R in for-
mula (4). Making these changes, we have the following:
THEOREM 1 Present Value of an Ordinary Annuity

1 - (1 + i)-n
PV = PMT (5)
i
where PV = present value of all payments
i = rate per period
n = number of periods
Note: Payments are made at the end of each period.
EXAMPLE 1 Present Value of an Annuity What is the present value of an annuity that pays
$200 per month for 5 years if money is worth 6% compounded monthly?
SOLUTION To solve this problem, use formula (5) with PMT = $200,
0.06
i = 12 = 0.005, and n = 12152 = 60:
1 - 11 + i2 -n
PV = PMT
i
1 - 11.0052 -60
= 200 Use a calculator.
0.005
= $10,345.11
Matched Problem 1 How much should you deposit in an account paying 8%
compounded quarterly in order to receive quarterly payments of $1,000 for the
next 4 years?
EXAMPLE 2 Retirement Planning Lincoln Benefit Life offered an ordinary annuity that earned
6.5% compounded annually. A person plans to make equal annual deposits into this
account for 25 years and then make 20 equal annual withdrawals of $25,000, reduc-
ing the balance in the account to zero. How much must be deposited annually to
accumulate sufficient funds to provide for these payments? How much total interest
is earned during this entire 45-year process?
SOLUTION This problem involves both future and present values. Figure 2 illus-
trates the flow of money into and out of the annuity.
Future value: Increasing balance Present value: Decreasing balance
FV 5 PV
Payments $PMT $PMT $PMT $25,000 $25,000 $25,000
Years 1 2 ??? 25 1 2 ??? 20

Figure 2
Since we are given the required withdrawals, we begin by finding the pres-
ent value necessary to provide for these withdrawals. Using formula (5) with
PMT = $25,000, i = 0.065, and n = 20, we have
1 - 11 + i2 -n
PV = PMT
i
1 - 11.0652 -20
= 25,000 Use a calculator.
0.065
= $275,462.68
Now we find the deposits that will produce a future value of $275,462.68 in
25 years. Using formula (7) from Section 2.3 with FV = $275,462.68, i = 0.065,
and n = 25, we have
i
PMT = FV
11 + i2 n - 1
0.065
= 275,462.68 Use a calculator.
11.0652 25 - 1
= $4,677.76
Thus, depositing $4,677.76 annually for 25 years will provide for 20 annual with-
drawals of $25,000. The interest earned during the entire 45-year process is
interest = 1total withdrawals2 - 1total deposits2
= 20125,0002 - 251$4,677.762
= $383,056
Matched Problem 2 Refer to Example 2. If $2,000 is deposited annually for
the first 25 years, how much can be withdrawn annually for the next 20 years?
Amortization
The present value formula for an ordinary annuity, formula (5), has another important
use. Suppose that you borrow $5,000 from a bank to buy a car and agree to repay the
loan in 36 equal monthly payments, including all interest due. If the bank charges 1%
per month on the unpaid balance (12% per year compounded monthly), how much
should each payment be to retire the total debt, including interest, in 36 months?
Actually, the bank has bought an annuity from you. The question is: If the bank
pays you $5,000 (present value) for an annuity paying them $PMT per month for
36 months at 12% interest compounded monthly, what are the monthly payments
(PMT)? (Note that the value of the annuity at the end of 36 months is zero.) To find
PMT, we have only to use formula (5) with PV = $5,000, i = 0.01, and n = 36:
1 - 11 + i2 -n
PV = PMT
i
1 - 11.012 -36
5,000 = PMT Solve for PMT and use a calculator.
0.01
PMT = $166.07 per month
At $166.07 per month, the car will be yours after 36 months. That is, you have amor-
tized the debt in 36 equal monthly payments. (Mort means “death”; you have “killed”
the loan in 36 months.) In general, amortizing a debt means that the debt is retired
in a given length of time by equal periodic payments that include compound inter-
est. We are interested in computing the equal periodic payments. Solving the present
value formula (5) for PMT in terms of the other variables, we obtain the following
amortization formula:
i
PMT = PV (6)
1 - 11 + i2 -n
Formula (6) is simply a variation of formula (5), and either formula can be used to
find the periodic payment PMT.
EXAMPLE 3 Monthly Payment and Total Interest on an Amortized Debt Assume that you
buy a TV for $800 and agree to pay for it in 18 equal monthly payments at 1 12%
interest per month on the unpaid balance.
(A) How much are your payments? (B) How much interest will you pay?
SOLUTION
(A) Use formula (5) or (6) with PV = $800, i = 0.015, n = 18, and solve for PMT:
i
PMT = PV
1 - 11 + i2 -n
0.015
1 - 11.0152 -18
= $51.04 per month
(B) Total interest paid = 1amount of all payments2 - 1initial loan2

= 181$51.042 - $800
= $118.72
Matched Problem 3 If you sell your car to someone for $2,400 and agree to
finance it at 1% per month on the unpaid balance, how much should you receive
each month to amortize the loan in 24 months? How much interest will you receive?

To purchase a home, a family plans to sign a mortgage of $70,000 at 8% on the
unpaid balance. Discuss the advantages and disadvantages of a 20-year mortgage as
opposed to a 30-year mortgage. Include a comparison of monthly payments and total
interest paid.
Amortization Schedules
What happens if you are amortizing a debt with equal periodic payments and later
decide to pay off the remainder of the debt in one lump-sum payment? This occurs
each time a home with an outstanding mortgage is sold. In order to understand what
happens in this situation, we must take a closer look at the amortization process. We be-
gin with an example that allows us to examine the effect each payment has on the debt.
EXAMPLE 4 Constructing an Amortization Schedule If you borrow $500 that you agree to
repay in six equal monthly payments at 1% interest per month on the unpaid bal-
ance, how much of each monthly payment is used for interest and how much is used
to reduce the unpaid balance?
SOLUTION First, we compute the required monthly payment using formula (5) or
(6). We choose formula (6) with PV = $500, i = 0.01, and n = 6:
i
PMT = PV
1 - 11 + i2 -n
0.01
1 - 11.012 -6
= $86.27 per month
At the end of the first month, the interest due is
$50010.012 = $5.00
The amortization payment is divided into two parts, payment of the interest due and
reduction of the unpaid balance (repayment of principal):
Monthly Interest Unpaid balance
payment due reduction
$86.27 = $5.00 + $81.27
The unpaid balance for the next month is
Previous Unpaid New
unpaid balance unpaid
balance reduction balance
$500.00 - $81.27 = $418.73
At the end of the second month, the interest due on the unpaid balance of $418.73 is
$418.7310.012 = $4.19
Thus, at the end of the second month, the monthly payment of $86.27 covers inter-
est and unpaid balance reduction as follows:
$86.27 = $4.19 + $82.08
and the unpaid balance for the third month is
$418.73 - $82.08 = $336.65
This process continues until all payments have been made and the unpaid balance is
reduced to zero. The calculations for each month are listed in Table 1, often referred
to as an amortization schedule.
Table 1 Amortization Schedule

Payment Unpaid Balance Unpaid
Number Payment Interest Reduction Balance
0 $500.00
1 $86.27 $5.00 $81.27 418.73
2 86.27 4.19 82.08 336.65
3 86.27 3.37 82.90 253.75
4 86.27 2.54 83.73 170.02
5 86.27 1.70 84.57 85.45
6 86.30 0.85 85.45 0.00
Totals $517.65 $17.65 $500.00
Matched Problem 4 Construct the amortization schedule for a $1,000 debt

that is to be amortized in six equal monthly payments at 1.25% interest per month
on the unpaid balance.
CONCEPTUAL INSIGHT
In Table 1, notice that the last payment had to be increased by $0.03 in order to
reduce the unpaid balance to zero. This small discrepancy is due to rounding the
monthly payment and the entries in the interest column to two decimal places.
Suppose that a family is making monthly payments on a home mortgage loan. If

the family decides to borrow money to make home improvements, they might take
out a home equity loan. The amount of the loan will depend on the equity in their
home, defined as the current net market value (the amount that would be received if
the home were sold, after subtracting all costs involved in selling the house) minus
the unpaid loan balance:
Equity = 1current net market value2 - 1unpaid loan balance2.
Similarly, if a family decides to sell the home they own and buy a more expensive
home, the equity in their current home will be an important factor in determining the
new home price that they can afford.
EXAMPLE 5 Equity in a Home A family purchased a home 10 years ago for $80,000. The
home was financed by paying 20% down and signing a 30-year mortgage at 9%
on the unpaid balance. The net market value of the house is now $120,000, and the
family wishes to sell the house. How much equity (to the nearest dollar) does the
family have in the house now after making 120 monthly payments?
SOLUTION How can we find the unpaid loan balance after 10 years or 120 monthly
payments? One way to proceed would be to construct an amortization schedule, but
this would require a table with 120 lines. Fortunately, there is an easier way. The
unpaid balance after 120 payments is the amount of the loan that can be paid off with
the remaining 240 monthly payments (20 remaining years on the loan). Since the
lending institution views a loan as an annuity that they bought from the family, the
unpaid balance of a loan with n remaining payments is the present value of that
annuity and can be computed by using formula (5). Since formula (5) requires
knowledge of the monthly payment, we compute PMT first using formula (6).
Step 1 Find the monthly payment:

i PV = 10.802 1 $80,0002 = $64,000
PMT = PV
1 - 11 + i2 -n i = 0.09
12 = 0.0075
0.0075
= 64,000 n = 121302 = 360
1 - 11.00752 -360
= $514.96 per month Use a calculator.
Step 2 Find the present value of a $514.96 per month, 20-year annuity:
1 - 11 + i2 -n PMT = $514.96
PV = PMT n = 121202 = 240
i
1 - 11.00752 -240 i = 0.09
12 = 0.0075
= 514.96 Use a calculator.
0.0075
= $57,235 Unpaid loan balance
Step 3 Find the equity:
equity = 1current net market value2 - 1unpaid loan balance2

= $120,000 - $57,235
= $62,765
So the equity in the home is $62,765. In other words, if the family sells the house for
$120,000 net, the family will walk away with $62,765 after paying off the unpaid
loan balance of $57,235.
Matched Problem 5 A couple purchased a home 20 years ago for $65,000.

The home was financed by paying 20% down and signing a 30-year mortgage at
8% on the unpaid balance. The net market value of the house is now $130,000, and
the couple wishes to sell the house. How much equity (to the nearest dollar) does
the couple have in the house now after making 240 monthly payments?
The unpaid loan balance in Example 5 may seem a surprisingly large amount to
owe after having made payments for 10 years, but long-term amortizations start out
with very small reductions in the unpaid balance. For example, the interest due at the
end of the very first period of the loan in Example 5 was
$64,00010.00752 = $480.00
The first monthly payment was divided as follows:
Unpaid
Monthly Interest balance
payment due reduction
$514.96 - $480.00 = $34.96
Only $34.96 was applied to the unpaid balance.

(A) A family has an $85,000, 30-year mortgage at 9.6% compounded monthly.
Show that the monthly payments are $720.94.
(B) Explain why the equation
85000 1 - 11.0082 -12130 - x2

y = 720.94
0.008
gives the unpaid balance of the loan after x years.
(C) Find the unpaid balance after 5 years, after 10 years, and after 15 years.
0 30
(D) When does the unpaid balance drop below half of the original $85,000?
0
(E) Solve part (D) using graphical approximation techniques on a graphing
Figure 3 calculator (see Fig. 3).
EXAMPLE 6 Automobile Financing You have negotiated a price of $25,200 for a new Bison
pickup truck. Now you must choose between 0% financing for 48 months or a
$3,000 rebate. If you choose the rebate, you can obtain a credit union loan for the
balance at 4.5% compounded monthly for 48 months. Which option should you
choose?
SOLUTION If you choose 0% financing, your monthly payment will be
25,200
PMT1 = = $525
48
If you choose the $3,000 rebate, and borrow $22,200 at 4.5% compounded monthly
for 48 months, the monthly payment is
ALL 2020 BISONS i
PMT2 = PV PV = $22,200
$3,000 Rebate 1 - 11 + i2 -n
or 0% APR for 4 years/48 months
0.00375 .045
= 22,200 i = 12 = 0.00375
1 - 1.00375 -48
= $506.24 n = 48
You should choose the $3,000 rebate. You will save 525 - 506.24 = $18.76
monthly or 48118.762 = $900.48 over the life of the loan.
Matched Problem 6 Which option should you choose if your credit union
raises its loan rate to 7.5% compounded monthly and all other data remain the
same?
EXAMPLE 7 Credit Cards The annual interest rate on a credit card is 18.99%. How long will it
take to pay off an unpaid balance of $847.29 if no new purchases are made and the
minimum payment of $20.00 is made each month?
SOLUTION It is necessary to make some simplifying assumptions because the
lengths of the billing cycles, the days on which payments are credited, and the
method for calculating interest are not specified. So we assume that there are 12
equal billing cycles per year, and that the $20 payments are credited at the end of
each cycle. With these assumptions, it is reasonable to use the present value formula
with PV = $847.29, PMT = $20.00, and i = 0.1899>12 in order to solve for n,

the number of payments:
1 - 11 + i2 -n
PV = PMT Multiply by i and divide by PMT.
i
i1PV>PMT2 = 1 - 11 + i2 -n Solve for 11 + i2 -n.
11 + i2 -n = 1 - i1PV>PMT2 Take the ln of both sides.
- n ln11 + i2 = ln11 - i1PV>PMT22 Solve for n.
ln11 - i1PV>PMT22
n = - Substitute i = 0.1899>12,
ln11 + i2
n ≈ 70.69 PV = $847.29, and PMT = $20.
We conclude that the unpaid balance will be paid off in 71 months.
Matched Problem 7 The annual interest rate on a credit card is 24.99%. How
long will it take to pay off an unpaid balance of $1,485.73 if no new purchases are
made and a $50.00 payment is made each month?
General Problem-Solving Strategy

After working the problems in Exercises 2.4, it is important to work the problems
in the Review Exercises. This will give you valuable experience in distinguish-
ing among the various types of problems we have considered in this chapter. It is
impossible to completely categorize all the problems you will encounter, but you
may find the following guidelines helpful in determining which of the four basic
formulas is involved in a particular problem. Be aware that some problems may
involve more than one of these formulas and others may not involve any of them.
SUMMARY Strategy for Solving Mathematics of Finance Problems

Step 1 Determine whether the problem involves a single payment or a sequence of
equal periodic payments. Simple and compound interest problems involve
a single present value and a single future value. Ordinary annuities may be
concerned with a present value or a future value but always involve a se-
quence of equal periodic payments.
Step 2 If a single payment is involved, determine whether simple or compound
interest is used. Often simple interest is used for durations of a year or less
and compound interest for longer periods.
Step 3 If a sequence of periodic payments is involved, determine whether the
payments are being made into an account that is increasing in value—a
future value problem—or the payments are being made out of an account
that is decreasing in value—a present value problem. Remember that
amortization problems always involve the present value of an ordinary
annuity.
Steps 1–3 will help you choose the correct formula for a problem, as indicated in
Figure 4. Then you must determine the values of the quantities in the formula that are
given in the problem and those that must be computed, and solve the problem.
Simple
interest A = P11 + rt2
Single
payment
Compound interest
(or continuous A = P11 + i2 n 1or A = Pert 2
compound interest)
Start
Future 11 + i2 n - 1
value of an FV = PMT
Sequence ordinary annuity i
of
payments Present 1 - 11 + i2 -n
value of an PV = PMT
ordinary annuity i
Figure 4 Selecting the correct formula for a problem
Exercises 2.4
W Skills Warm-up Exercises B In Problems 15–22, use formula (5) or (6) to solve each problem.
In Problems 1–6, find the sum of the finite geometric series 15. n = 30; i = 0.04; PMT = $200; PV = ?
a + ar + ar + g + ar . Write the answer as a quotient of
2 n-1
16. n = 24; i = 0.015; PMT = $250; PV = ?
integers. (If necessary, review Section C.2).
17. PV = $40,000; n = 96; i = 0.0075; PMT = ?
1 1 1 1
1. 1 + + + + g+ 8
2 4 8 2 18. PV = $25,000; n = 48; i = 0.0095; PMT = ?
1 1 1 1 19. PV = $5,000; i = 0.01; PMT = $200; n = ?
2. 1 + + + + g+ 7
5 25 125 5
3 3 3 20. PV = $12,000; i = 0.0125; PMT = $750; n = ?
3. 30 + 3 + + + g+
10 100 1,000,000 21. PV = $9,000; PMT = $600; n = 20; i = ? (Round answer
1 to three decimal places.)
4. 10,000 + 1,000 + 100 + 10 + g +
10,000 22. PV = $12,000; PMT = $400; n = 40; i = ? (Round
1 1 1 1 answer to three decimal places.)
5. 1 - + - + g+ 6
3 9 27 3 C
1 1 1 1 1
6. 1 - + - + - 23. Explain what is meant by the present value of an ordinary
10 100 1,000 10,000 100,000
annuity.
A In Problems 7–14, find i (the rate per period) and n (the number 24. Solve the present value formula (5) for n.
of periods) for each loan at the given annual rate.
25. Explain how an ordinary annuity is involved when you take
7. Monthly payments of $87.65 are made for 7 years to repay a out an auto loan from a bank.
loan at 5.4% compounded monthly.
26. Explain why the last payment in an amortization schedule
8. Semiannual payments of $3,200 are made for 12 years to might differ from the other payments.
repay a loan at 9.9% compounded semiannually.
9. Quarterly payments of $975 are made for 10 years to repay a
loan at 9.9% compounded quarterly. Applications
10. Annual payments of $1,045 are made for 5 years to repay a 27. A bank offers a 9-year ordinary annuity with a guaranteed rate
loan at 4.75% compounded annually. of 3.70% compounded annually. How much should you pay
11. Semiannual payments of $1,250 are made for 25 years to for one of these annuities if you want to receive payments of
repay a loan at 6.75% compounded semiannually. $8,000 annually over the 9-year period?
12. Quarterly payments of $610 are made for 6 years to repay 28. A bank offers a 5-year ordinary annuity with a guaranteed
loan at 8.24% compounded quarterly. rate of 3.30% compounded annually. How much should
you pay for one of these annuities if you want to receive
13. Annual payments of $5,195 are made for 9 years to repay a
payments of $5,000 annually over the 5-year period?
loan at 5.48% compounded annually.
29. E-Loan, an online lending service, offers a 36-month auto
14. Monthly payments of $433 are made for 3 years to repay a
loan at 7.56% compounded monthly to applicants with good
loan at 10.8% compounded monthly.
credit ratings. If you have a good credit rating and can afford
monthly payments of $350, how much can you borrow from 40. You want to purchase an automobile for $28,500. The dealer
E-Loan? What is the total interest you will pay for this loan? offers you 0% financing for 60 months or a $6,000 rebate.
You can obtain 6.2% financing for 60 months at the local
30. E-Loan offers a 36-month auto loan at 9.84% compounded
bank. Which option should you choose? Explain.
monthly to applicants with fair credit ratings. If you have a
fair credit rating and can afford monthly payments of $350, 41. A sailboat costs $35,000. You pay 20% down and amortize
how much can you borrow from E-Loan? What is the total the rest with equal monthly payments over a 12-year period.
interest you will pay for this loan? If you must pay 8.75% compounded monthly, what is your
monthly payment? How much interest will you pay?
31. If you buy a computer directly from the manufacturer for
$2,500 and agree to repay it in 48 equal installments at 1.25% 42. A recreational vehicle costs $80,000. You pay 10% down and
interest per month on the unpaid balance, how much are your amortize the rest with equal monthly payments over a 7-year
monthly payments? How much total interest will be paid? period. If you pay 9.25% compounded monthly, what is your
monthly payment? How much interest will you pay?
32. If you buy a computer directly from the manufacturer for
$3,500 and agree to repay it in 60 equal installments at 1.75% 43. Construct the amortization schedule for a $5,000 debt that
interest per month on the unpaid balance, how much are your is to be amortized in eight equal quarterly payments at 2.8%
monthly payments? How much total interest will be paid? interest per quarter on the unpaid balance.
In Problems 33–36, assume that no new purchases are made with 44. Construct the amortization schedule for a $10,000 debt that is
the credit card. to be amortized in six equal quarterly payments at 2.6% inter-
est per quarter on the unpaid balance.
33. The annual interest rate on a credit card is 16.99%. If a pay-
ment of $100.00 is made each month, how long will it take to 45. A woman borrows $6,000 at 9% compounded monthly,
pay off an unpaid balance of $2,487.56? which is to be amortized over 3 years in equal monthly pay-
ments. For tax purposes, she needs to know the amount of
34. The annual interest rate on a credit card is 24.99%. If a pay- interest paid during each year of the loan. Find the interest
ment of $100.00 is made each month, how long will it take to paid during the first year, the second year, and the third year
pay off an unpaid balance of $2,487.56? of the loan. [Hint: Find the unpaid balance after 12 payments
35. The annual interest rate on a credit card is 14.99%. If the and after 24 payments.]
minimum payment of $20 is made each month, how long will 46. A man establishes an annuity for retirement by deposit-
it take to pay off an unpaid balance of $937.14? ing $50,000 into an account that pays 7.2% compounded
monthly. Equal monthly withdrawals will be made each
36. The annual interest rate on a credit card is 22.99%. If the
month for 5 years, at which time the account will have a
minimum payment of $25 is made each month, how long will
zero balance. Each year taxes must be paid on the interest
it take to pay off an unpaid balance of $860.22?
earned by the account during that year. How much interest
Problems 37 and 38 refer to the following ads. was earned during the first year? [Hint: The amount in the
account at the end of the first year is the present value of a
37. The ad for a Bison sedan claims that a monthly payment of
4-year annuity.]
$299 constitutes 0% financing. Explain why that is false.
Find the annual interest rate compounded monthly that is ac- 47. Some friends tell you that they paid £2,400 down on a new
tually being charged for financing $17,485 with 72 monthly car and are to pay £299 per month for 48 months. If interest
payments of $299. is 3.9% compounded monthly, what was the selling price of
the car? How much interest will they pay in 48 months?
2020 BISON SEDAN 2020 BISON SUV 48. A couple is thinking about buying a new car costing
Zero down – 0% financing Zero down – 0% financing
$299 per month* $399 per month* £20,000. The couple must pay 15% down, and the rest is
to be amortized over 48 months in equal monthly payments.
If money costs 2.9% compounded monthly, what will the
monthly payment be? How much total interest will be paid
over 48 months?
Buy for $17,485.
$17 485 Buy for $23,997.
$23 997
* Bison sedan, 0% down, 0% for 72 months * Bison SUV, 0% down, 0% for 72 months
49. A student receives a federally backed student loan of $6,000
at 3.5% interest compounded monthly. After finishing college
38. The ad for a Bison SUV claims that a monthly payment of in 2 years, the student must amortize the loan in the next
$399 constitutes 0% financing. Explain why that is false. 4 years by making equal monthly payments. What will the
Find the annual interest rate compounded monthly that is ac- payments be and what total interest will the student pay?
tually being charged for financing $23,997 with 72 monthly [Hint: This is a two-part problem. First, find the amount of
payments of $399. the debt at the end of the first 2 years; then amortize this
amount over the next 4 years.]
39. You want to purchase an automobile for $27,300. The dealer
offers you 0% financing for 60 months or a $5,000 rebate. 50. A person establishes a sinking fund for retirement by contrib-
You can obtain 6.3% financing for 60 months at the local uting $7,500 per year at the end of each year for 20 years. For
bank. Which option should you choose? Explain. the next 20 years, equal yearly payments are withdrawn, at
the end of which time the account will have a zero balance. If
money is worth 9% compounded annually, what yearly pay- monthly withdrawals of $1,500 for the next 20 years, re-
ments will the person receive for the last 20 years? ducing the balance to zero. How much should be depos-
ited each month for the first 15 years? What is the total
51. A family has a $150,000, 30-year mortgage at 6.1% com-
interest earned during this 35-year process?
pounded monthly. Find the monthly payment. Also find the
unpaid balance after (B) If the person makes monthly deposits of $1,000 for the
first 15 years, how much can be withdrawn monthly for
(A) 10 years (B) 20 years
the next 20 years?
(C) 25 years
59. A couple wishes to borrow money using the equity in their
52. A family has a $210,000, 20-year mortgage at 6.75% com- home for collateral. A loan company will loan the couple up to
pounded monthly. Find the monthly payment. Also find the 70% of their equity. The couple purchased the home 12 years
unpaid balance after ago for $179,000. The home was financed by paying 20%
down and signing a 30-year mortgage at 8.4% on the unpaid
(A) 5 years (B) 10 years
balance. Equal monthly payments were made to amortize
(C) 15 years the loan over the 30-year period. The net market value of the
house is now $215,000. After making the 144th payment, the
53. A family has a $129,000, 20-year mortgage at 7.2% com-
couple applied to the loan company for the maximum loan.
pounded monthly.
How much (to the nearest dollar) will the couple receive?
(A) Find the monthly payment and the total interest paid.
60. A person purchased a house 10 years ago for $160,000.
(B) Suppose the family decides to add an extra $102.41 to The house was financed by paying 20% down and signing
its mortgage payment each month starting with the very a 30-year mortgage at 7.75% on the unpaid balance. Equal
first payment. How long will it take the family to pay off monthly payments were made to amortize the loan over a 30-
the mortgage? How much interest will be saved? year period. The owner now (after the 120th payment) wishes
to refinance the house due to a need for additional cash. If
54. At the time they retire, a couple has $200,000 in an account
the loan company agrees to a new 30-year mortgage of 80%
that pays 8.4% compounded monthly.
of the new appraised value of the house, which is $225,000,
(A) If the couple decides to withdraw equal monthly pay- how much cash (to the nearest dollar) will the owner receive
ments for 10 years, at the end of which time the account after repaying the balance of the original mortgage?
will have a zero balance, how much should the couple
61. A person purchased a $145,000 home 10 years ago by paying
withdraw each month?
20% down and signing a 30-year mortgage at 7.9% com-
(B) If the couple decides to withdraw $3,000 a month until pounded monthly. Interest rates have dropped and the owner
the balance in the account is zero, how many withdraw- wants to refinance the unpaid balance by signing a new
als can the couple make? 20-year mortgage at 5.5% compounded monthly. How much
interest will refinancing save?
55. An ordinary annuity that earns 7.5% compounded monthly
has a current balance of $500,000. The owner of the account 62. A person purchased a $200,000 home 20 years ago by paying
is about to retire and has to decide how much to withdraw 20% down and signing a 30-year mortgage at 13.2% com-
from the account each month. Find the number of withdraw- pounded monthly. Interest rates have dropped and the owner
als under each of the following options: wants to refinance the unpaid balance by signing a new
10-year mortgage at 8.2% compounded monthly. How much
(A) $5,000 monthly (B) $4,000 monthly interest will refinancing save?
(C) $3,000 monthly 63. Discuss the similarities and differences in the graphs of
56. Refer to Problem 55. If the account owner decides to unpaid balance as a function of time for 30-year mortgages
withdraw $3,000 monthly, how much is in the account after of $50,000, $75,000, and $100,000, respectively, each at 9%
10 years? After 20 years? After 30 years? compounded monthly (see the figure). Include computations
of the monthly payment and total interest paid in each case.
57. An ordinary annuity pays 7.44% compounded monthly.
(A) A person deposits $100 monthly for 30 years and then 100,000
makes equal monthly withdrawals for the next 15 years,
reducing the balance to zero. What are the monthly with- 80,000
Unpaid balance ($)
drawals? How much interest is earned during the entire

45-year process? 60,000
(B) If the person wants to make withdrawals of $2,000 per 40,000

month for the last 15 years, how much must be deposited
20,000
monthly for the first 30 years?
58. An ordinary annuity pays 6.48% compounded monthly. 0 t
0 6 12 18 24 30
Time (years)
(A) A person wants to make equal monthly deposits into
the account for 15 years in order to then make equal Figure for 63
64. Discuss the similarities and differences in the graphs of 67. The owner of a small business has received two offers of
unpaid balance as a function of time for 30-year mortgages of purchase. The first prospective buyer offers to pay the owner
$60,000 at rates of 7%, 10%, and 13%, respectively (see the $100,000 in cash now. The second offers to pay the owner
figure). Include computations of the monthly payment and $10,000 now and monthly payments of $1,200 for 10 years.
total interest paid in each case. In effect, the second buyer is asking the owner for a $90,000
loan. If the owner accepts the second offer, what annual
60,000
nominal compounding rate will the owner receive for financ-
ing this purchase?
50,000
Unpaid balance ($)
68. At the time they retire, a couple has $200,000 invested

40,000
in an annuity. The couple can take the entire amount in a
30,000 single payment, or receive monthly payments of $2,000 for
15 years. If the couple elects to receive the monthly pay-
20,000
ments, what annual nominal compounding rate will the
10,000 couple earn on the money invested in the annuity?
0 t
0 6 12 18 24 30
Time (years) Answers to Matched Problems
1. $13,577.71 2. $10,688.87
In Problems 65–68, use graphical approximation techniques or an
3. PMT = $112.98>mo; total interest = $311.52
equation solver to approximate the desired interest rate. Express
4. Unpaid
each answer as a percentage, correct to two decimal places.
Payment Balance Unpaid
65. A discount electronics store offers to let you pay for a $1,000 Number Payment Interest Reduction Balance
stereo in 12 equal $90 installments. The store claims that
0 $1,000.00
since you repay $1,080 in 1 year, the $80 finance charge rep-
1 $174.03 $12.50 $161.53 838.47
resents an 8% annual rate. This would be true if you repaid
2 174.03 10.48 163.55 674.92
the loan in a single payment at the end of the year. But since
you start repayment after 1 month, this is an amortized loan, 3 174.03 8.44 165.59 509.33
and 8% is not the correct rate. What is the annual nominal 4 174.03 6.37 167.66 341.67
compounding rate for this loan? 5 174.03 4.27 169.76 171.91
6 174.06 2.15 171.91 0.00
66. A $2,000 computer can be financed by paying $100 per Totals $1,044.21 $44.21 $1,000.00
month for 2 years. What is the annual nominal compounding
rate for this loan? 5. $98,551 6. Choose the 0% financing. 7. 47 months

2.1 Simple Interest EXAMPLES
• Interest is the fee paid for the use of a sum of money P, called the principal. Simple interest is given by
I = Prt
where I = interest
P = principal
r = annual simple interest rate 1written as a decimal2
t = time in years
• If a principal P (present value) is borrowed, then the amount A (future value) is the total of the princi-
Ex. 1, p. 106
pal and the interest:
Ex. 2, p. 106
A = P + Prt Ex. 3, p. 107
= P11 + rt2 Ex. 4, p. 107
Ex. 5, p. 108
• The average daily balance method is a common method for calculating the interest owed on a credit Ex. 6, p. 109
card. The formula I = Prt is used, but a daily balance is calculated for each day of the billing cycle, and
P is the average of those daily balances.
2.2 Compound and Continuous Compound Interest

• Compound interest is interest paid on the principal plus reinvested interest. The future and present Ex. 1, p. 114
values are related by
A = P11 + i2 n
where i = r>m and
A = amount or future value
P = principal or present value
r = annual nominal rate 1or just rate2
i = rate per compounding period
n = total number of compounding periods
• If a principal P is invested at an annual rate r earning continuous compound interest, then the amount Ex. 2, p. 116
A after t years is given by Ex. 3, p. 117
Ex. 4, p. 118
A = Pen
• The growth time of an investment is the time it takes for a given principal to grow to a particular Ex. 5, p. 119
amount. Three methods for finding the growth time are as follows:
1. Use logarithms and a calculator.
2. Use graphical approximation on a graphing calculator.
3. Use an equation solver on a graphing calculator or a computer.
• The annual percentage yield (APY; also called the effective rate or true interest rate) is the simple
interest rate that would earn the same amount as a given annual rate for which interest is compounded.
• If a principal is invested at the annual rate r compounded m times a year, then the annual percentage Ex. 6, p. 121
yield is given by
r m
APY = a1 + b - 1
m
• If a principal is invested at the annual rate r compounded continuously, then the annual percentage yield Ex. 7, p. 121
is given by
APY = er - 1
• A zero coupon bond is a bond that is sold now at a discount and will pay its face value at some time in
the future when it matures.
2.3 Future Value of an Annuity; Sinking Funds
• An annuity is any sequence of equal periodic payments. If payments are made at the end of each time Ex. 1, p. 128
interval, then the annuity is called an ordinary annuity. The amount, or future value, of an annuity is Ex. 3, p. 131
the sum of all payments plus all interest earned and is given by Ex. 4, p. 131
11 + i2 n - 1
FV = PMT
i
where FV = future value 1amount2
i = rate per period
n = number of payments 1periods2
• A balance sheet is a table that shows the interest and balance for each payment of an annuity.
• An account that is established to accumulate funds to meet future obligations or debts is called a sinking Ex. 2, p. 130
fund. The sinking fund payment can be found by solving the future value formula for PMT:
i
PMT = FV
11 + i2 n - 1
2.4 Present Value of an Annuity; Amortization

• If equal payments are made from an account until the amount in the account is 0, the payment and the Ex. 1, p. 136
present value are related by the following formula: Ex. 2, p. 136
1 - 11 + i2 -n
PV = PMT
i
where PV = present value of all payments
i = rate per period
n = number of periods
• Amortizing a debt means that the debt is retired in a given length of time by equal periodic payments Ex. 3, p. 137
that include compound interest. Solving the present value formula for the payment gives us the amorti- Ex. 6, p. 141
zation formula: Ex. 7, p. 141
i
PMT = PV
1 - 11 + i2 -n
• An amortization schedule is a table that shows the interest due and the balance reduction for each pay- Ex. 4, p. 138
ment of a loan.
• The equity in a property is the difference between the current net market value and the unpaid loan bal- Ex. 5, p. 139
ance. The unpaid balance of a loan with n remaining payments is given by the present value formula.
• A strategy for solving problems in the mathematics of finance is presented on page 142.
Review Exercises
Work through all the problems in this chapter review and check B In Problems 9 and 10, find the indicated quantity, given
your answers in the back of the book. Answers to all review prob-
lems are there along with section numbers in italics to indicate 11 + i2 n - 1
where each type of problem is discussed. Where weaknesses show FV = PMT
i
9. FV = $10,000; PMT = ?; i = 0.016; n = 54
A In Problems 1–4, find the indicated quantity, given
A = P11 + rt2. 10. FV = ?; PMT = $1,200; i = 0.004; n = 72
1. A = ?; P = $200; r = 9%; t = 8 months In Problems 11 and 12, find the indicated quantity, given
2. A = $900; P = ?; r = 14%; t = 3 month
1 - 11 + i2 -n
3. A = $312; P = $250; r = 7%; t = ? PV = PMT
i
4. A = $3,120; P = $3,000; r = ?; t = 8 months 11. FV = ?; PMT = $3,500; i = 0.03; n = 14
In Problems 5 and 6, find the indicated quantity, given 12. FV = $9,000; PMT = ?; i = 0.0065; n = 16
A = P11 + i2 n.
C 13. Solve the equation 2,500 = 1,00011.062 n for n to the near-
5. A = ?; P = $1,400; i = 0.004; n = 30 est integer using:
6. A = $6,000; P = ?; i = 0.006; n = 62 (A) Logarithms
rt
In Problems 7 and 8, find the indicated quantity, given A = Pe . (B) Graphical approximation techniques or an equation
7. A = ?; P = $5,400; r = 5.8%; t = 2 years solver on a graphing calculator
8. A = 45,000; P = ?; r = 9.4%; t = 72 months
14. Solve the equation The unpaid balance at the start of a 30-day billing cycle was
$1,672.18. A purchase of $265.12 was made on day 8 and a
11.012 n - 1 payment of $250 was credited to the account on day 20.
5,000 = 100 Find the unpaid balance at the end of the billing cycle.
0.01
(Use a 360-day year.)
for n to the nearest integer using:
26. What will a $23,000 car cost (to the nearest dollar) 5 years
(A) Logarithms
from now if the inflation rate over that period averages 5%
(B) Graphical approximation techniques or an equation compounded annually?
solver on a graphing calculator.
27. What would the $23,000 car in Problem 26 have cost (to
the nearest dollar) 5 years ago if the inflation rate over that
period had averaged 5% compounded annually?
Applications 28. A loan of $2,500 was repaid at the end of 10 months with
a check for $2,812.50. What annual rate of interest was
Find all dollar amounts correct to the nearest cent. When an inter- charged?
est rate is requested as an answer, express the rate as a percent- 29. You want to purchase an automobile for $21,600. The dealer
age, correct to two decimal places. offers you 0% financing for 48 months or a $3,000 rebate.
15. You need to borrow $4,000 for 9 months. The local bank is You can obtain 4.8% financing for 48 months at the local
willing to supply the money at 9% simple interest. Calculate bank. Which option should you choose? Explain.
the amount of interest that will be paid on the loan for the stated 30. Find the annual percentage yield on a bond earning 6.25% if
period and the total amount required at the end of the period. interest is compounded
16. Grandparents deposited $6,000 into a grandchild’s account
(A) monthly.
toward a college education. How much money (to the nearest
dollar) will be in the account 17 years from now if the ac- (B) continuously.
count earns 7% compounded monthly?
31. You have $5,000 toward the purchase of a boat that will cost
17. How much should you pay for a corporate bond paying 6.6% $6,000. How long will it take the $5,000 to grow to $6,000 if
compounded monthly in order to have $25,000 in 10 years? it is invested at 9% compounded quarterly? (Round up to the
next-higher quarter if not exact.)
18. An investment account pays 5.4% compounded annually.
Construct a balance sheet showing the interest earned during 32. How long will it take money to double if it is invested at 6%
each year and the balance at the end of each year for 4 years if compounded monthly? 9% compounded monthly? (Round up
to the next-higher month if not exact.)
(A) A single deposit of $400 is made at the beginning of the
first year. 33. Starting on his 21st birthday, and continuing on every birth-
day up to and including his 65th, John deposits $2,000 a year
(B) Four deposits of $100 are made at the end of each year.
into an IRA. How much (to the nearest dollar) will be in the
19. One investment pays 13% simple interest and another 9% account on John’s 65th birthday, if the account earns:
compounded annually. Which investment would you choose?
(A) 7% compounded annually?
Why?
(B) 11% compounded annually?
20. A $10,000 retirement account is left to earn interest at 7%
compounded daily. How much money will be in the account 34. If you just sold a stock for $17,388.17 (net) that cost you
40 years from now when the owner reaches 65? (Use a 365- $12,903.28 (net) 3 years ago, what annual compound rate of
day year and round answer to the nearest dollar.) return did you make on your investment?
21. A couple wishes to have $40,000 in 6 years for the down pay- 35. The table shows the fees for refund anticipation loans
ment on a house. At what rate of interest compounded continu- (RALs) offered by an online tax preparation firm. Find the
ously must $25,000 be invested now to accomplish this goal? annual rate of interest for each of the following loans.
Assume a 360-day year.
22. Which is the better investment and why: 9% compounded
quarterly or 9.25% compounded annually? (A) A $400 RAL paid back in 15 days
23. What is the value of an ordinary annuity at the end of 8 years (B) A $1,800 RAL paid back in 21 days
if $200 per month is deposited into an account earning 7.2%
RAL Amount RAL Fee
compounded monthly? How much of this value is interest?
$10–$500 $29.00
24. A payday lender charges $60 for a loan of $500 for 15 days. $501–$1,000 $39.00
Find the annual interest rate. (Use a 360-day year.) $1,001–$1,500 $49.00
25. The annual interest rate on a credit card is 25.74% and $1,501–$2,000 $69.00
interest is calculated by the average daily balance method. $2,001–$5,000 $82.00
36. Lincoln Benefit Life offered an annuity that pays 5.5% com- 48. A company establishes a sinking fund for plant retooling in
pounded monthly. What equal monthly deposit should be 6 years at an estimated cost of $850,000. How much should
made into this annuity in order to have $50,000 in 5 years? be invested semiannually into an account paying 8.76%
compounded semiannually? How much interest will the ac-
37. A person wants to establish an annuity for retirement
count earn in the 6 years?
purposes. He wants to make quarterly deposits for 20 years
so that he can then make quarterly withdrawals of $5,000 49. What is the annual nominal rate compounded monthly for a
for 10 years. The annuity earns 7.32% interest compounded CD that has an annual percentage yield of 2.50%?
quarterly.
50. If you buy a 13-week T-bill with a maturity value of $5,000
(A) How much will have to be in the account at the time he for $4,922.15 from the U.S. Treasury Department, what
retires? annual interest rate will you earn?
(B) How much should be deposited each quarter for 20 years 51. In order to save enough money for the down payment on a
in order to accumulate the required amount? condominium, a young couple deposits $200 each month
into an account that pays 7.02% interest compounded
(C) What is the total amount of interest earned during the
monthly. If the couple needs $10,000 for a down payment,
30-year period?
how many deposits will the couple have to make?
38. If you borrow $4,000 from an online lending firm for the
52. A business borrows $80,000 at 9.42% interest compounded
purchase of a computer and agree to repay it in 48 equal in-
monthly for 8 years.
stallments at 0.9% interest per month on the unpaid balance,
how much are your monthly payments? How much total (A) What is the monthly payment?
interest will be paid?
(B) What is the unpaid balance at the end of the first year?
39. A company decides to establish a sinking fund to replace
(C) How much interest was paid during the first year?
a piece of equipment in 6 years at an estimated cost of
$50,000. To accomplish this, they decide to make fixed 53. You unexpectedly inherit $10,000 just after you have made
monthly payments into an account that pays 6.12% com- the 72nd monthly payment on a 30-year mortgage of
pounded monthly. How much should each payment be? $60,000 at 8.2% compounded monthly. Discuss the relative
merits of using the inheritance to reduce the principal of the
40. How long will it take money to double if it is invested at
loan or to buy a certificate of deposit paying 7% compound-
7.5% compounded daily? 7.5% compounded annually?
ed monthly.
41. A student receives a student loan for $8,000 at 5.5% interest
54. Your parents are considering a $75,000, 30-year mortgage
compounded monthly to help her finish the last 1.5 years of
to purchase a new home. The bank at which they have done
college. Starting 1 year after finishing college, the student
business for many years offers a rate of 7.54% compounded
must amortize the loan in the next 5 years by making equal
monthly. A competitor is offering 6.87% compounded
monthly payments. What will the payments be and what
monthly. Would it be worthwhile for your parents to switch
total interest will the student pay?
banks? Explain.
42. If you invest $5,650 in an account paying 8.65% compound-
55. How much should a $5,000 face value zero coupon bond,
ed continuously, how much money will be in the account at
maturing in 5 years, be sold for now, if its rate of return is to
the end of 10 years?
be 5.6% compounded annually?
43. A company makes a payment of $1,200 each month into
56. If you pay $5,695 for a $10,000 face value zero coupon bond
a sinking fund that earns 6% compounded monthly. Use
that matures in 10 years, what is your annual compound rate
graphical approximation techniques on a graphing calculator
of return?
to determine when the fund will be worth $100,000.
57. If an investor wants to earn an annual interest rate of 6.4%
44. A couple has a $50,000, 20-year mortgage at 9% compound-
on a 26-week T-bill with a maturity value of $5,000, how
ed monthly. Use graphical approximation techniques on a
much should the investor pay for the T-bill?
graphing calculator to determine when the unpaid balance
will drop below $10,000. 58. Two years ago you borrowed $10,000 at 12% interest com-
pounded monthly, which was to be amortized over 5 years.
45. A loan company advertises in the paper that you will pay
Now you have acquired some additional funds and decide
only 8¢ a day for each $100 borrowed. What annual rate of
that you want to pay off this loan. What is the unpaid bal-
interest are they charging? (Use a 360-day year.)
ance after making equal monthly payments for 2 years?
46. Construct the amortization schedule for a $1,000 debt that
59. What annual nominal rate compounded monthly has the same
is to be amortized in four equal quarterly payments at 2.5%
annual percentage yield as 7.28% compounded quarterly?
interest per quarter on the unpaid balance.
60. (A) A man deposits $2,000 in an IRA on his 21st birthday
47. You can afford monthly deposits of only $300 into an ac-
and on each subsequent birthday up to, and including,
count that pays 7.98% compounded monthly. How long
his 29th (nine deposits in all). The account earns 8%
will it be until you will have $9,000 to purchase a used car?
compounded annually. If he leaves the money in the
(Round to the next-higher month if not exact.)
account without making any more deposits, how much $83,000. The home was financed by paying 20% down and
will he have on his 65th birthday, assuming the account signing a 30-year mortgage at 8.4% for the balance. Equal
continues to earn the same rate of interest? monthly payments were made to amortize the loan over
the 30-year period. The market value of the house is now
(B) How much would be in the account (to the nearest dol-
$95,000. After making the 96th payment, the family applied
lar) on his 65th birthday if he had started the deposits
to the loan company for the maximum loan. How much (to
on his 30th birthday and continued making deposits on
the nearest dollar) will the family receive?
each birthday until (and including) his 65th birthday?
64. A $600 stereo is financed for 6 months by making monthly
61. A promissory note will pay $27,000 at maturity 10 years
payments of $110. What is the annual nominal compounding
from now. How much money should you be willing to pay
rate for this loan?
now if money is worth 5.5% compounded continuously?
65. A person deposits $2,000 each year for 25 years into an IRA.
62. In a new housing development, the houses are selling for
When she retires immediately after making the 25th deposit,
$100,000 and require a 20% down payment. The buyer
the IRA is worth $220,000.
is given a choice of 30-year or 15-year financing, both at
7.68% compounded monthly. (A) Find the interest rate earned by the IRA over the
25-year period leading up to retirement.
(A) What is the monthly payment for the 30-year choice?
For the 15-year choice? (B) Assume that the IRA continues to earn the interest rate
found in part (A). How long can the retiree withdraw
(B) What is the unpaid balance after 10 years for the 30-
$30,000 per year? How long can she withdraw $24,000
year choice? For the 15-year choice?
per year?
63. A loan company will loan up to 60% of the equity in a
home. A family purchased their home 8 years ago for
3 Systems of Linear
Equations; Matrices
3.1 Review: Systems of Linear
Equations in
Introduction
Two Variables Traffic congestion in urban areas is a major problem. Traffic control by on-
3.2 Systems of Linear ramp metering, traffic signals, and message signs is one approach to reducing
Equations and congestion. Traffic control by vehicle-to-vehicle communication shows promise
Augmented Matrices for the future. The mathematical prerequisite for solving problems in traffic flow
is the ability to solve systems of linear equations (see Problems 97 and 98 in
3.3 Gauss–Jordan Elimination
Section 3.3).
3.4 Matrices: Basic In Chapter 3 we discuss methods for solving systems of two linear equations
Operations by hand calculation. We also introduce matrix methods, including Gauss–
3.5 Inverse of a Jordan elimination, that can be used to solve systems with many equations on a
Square Matrix calculator or computer. We consider a number of important applications, includ-
ing resource allocation, production scheduling, and economic planning.
3.6 Matrix Equations
and Systems of
Linear Equations
3.7 Leontief Input–Output
Analysis
152
SECTION 3.1 Review: Systems of Linear Equations in Two Variables 153
3.1 Review: Systems of Linear Equations in Two Variables

■■ Systems of Linear Systems of Linear Equations in Two Variables
To establish basic concepts, let’s consider the following simple example: If 2 adult
■■ Graphing tickets and 1 child ticket cost $32, and if 1 adult ticket and 3 child tickets cost $36,
■■ Substitution what is the price of each?
■■ Elimination by Addition Let: x = price of adult ticket
■■ Applications y = price of child ticket
Then: 2x + y = 32
x + 3y = 36
Now we have a system of two linear equations in two variables. It is easy to find
ordered pairs (x, y) that satisfy one or the other of these equations. For example, the
ordered pair (16, 0) satisfies the first equation but not the second, and the ordered pair
(24, 4) satisfies the second but not the first. To solve this system, we must find all
ordered pairs of real numbers that satisfy both equations at the same time. In general,
we have the following definition:
DEFINITION Systems of Two Linear Equations in Two Variables

Given the linear system
ax + by = h
cx + dy = k
where a, b, c, d, h, and k are real constants, a pair of numbers x = x 0 and y = y0
[also written as an ordered pair 1x 0, y0 2] is a solution of this system if each equation
is satisfied by the pair. The set of all such ordered pairs is called the solution set for
the system. To solve a system is to find its solution set.
We will consider three methods of solving such systems: graphing, substitu-

tion, and elimination by addition. Each method has its advantages, depending on the
situation.
Graphing
Recall that the graph of a line is a graph of all the ordered pairs that satisfy the equa-
tion of the line. To solve the ticket problem by graphing, we graph both equations in
the same coordinate system. The coordinates of any points that the graphs have in
common must be solutions to the system since they satisfy both equations.
EXAMPLE 1 Solving a System by Graphing Solve the ticket problem by graphing:

2x + y = 32
x + 3y = 36
SOLUTION An easy way to find two distinct points on the first line is to find the
x and y intercepts. Substitute y = 0 to find the x intercept (2x = 32, so x = 16),
and substitute x = 0 to find the y intercept (y = 32). Then draw the line through
154 CHAPTER 3 Systems of Linear Equations; Matrices
Reminder (16, 0) and (0, 32). After graphing both lines in the same coordinate system (Fig. 1),
Recall that we may graph a line by estimate the coordinates of the intersection point:
finding its intercepts (Section A.2). y
For the equation 2x + y = 32:
If x = 0, then y = 32.
40
If y = 0, then x = 16.
This gives the points (0, 32) and x = $12 Adult ticket
(16, 0), which uniquely define the line. 20
y = $8 Child ticket
(12, 8)
x
0 20 40
2x 1 y 5 32 x 1 3y 5 36
Figure 1
2x + y = 32 x + 3y = 36
CHECK 21122 + 8 ≟ 32 12 + 3182 ≟ 36 Check that (12, 8) satisfies
✓ ✓
32 = 32 36 = 36 each of the original equations.
Matched Problem 1 Solve by graphing and check:

2x - y = - 3
x + 2y = - 4
It is clear that Example 1 has exactly one solution since the lines have exactly
one point in common. In general, lines in a rectangular coordinate system are related
to each other in one of the three ways illustrated in the next example.
EXAMPLE 2 Solving a System by Graphing Solve each of the following systems by graphing:
(A) x - 2y = 2 (B) x + 2y = - 4 (C) 2x + 4y = 8
x + y = 5 2x + 4y = 8 x + 2y = 4
SOLUTION
(A) y (B) y (C) y
5 5 5
(4, 1)
x x 5
x
25 0 5 25 0 5 25 0
x54
y51
25 25 25
Intersection at one point Lines are parallel each( Lines coincide—infinite

only—exactly one solution has slope 22
1
)
2 —no solutions
number of solutions
Matched Problem 2 Solve each of the following systems by graphing:

(A) x + y = 4 (B) 6x - 3y = 9 (C) 2x - y = 4
2x - y = 2 2x - y = 3 6x - 3y = - 18
We introduce some terms that describe the different types of solutions to systems
of equations.
DEFINITION Systems of Linear Equations: Basic Terms

A system of linear equations is consistent if it has one or more solutions and
inconsistent if no solutions exist. Furthermore, a consistent system is said to be
independent if it has exactly one solution (often referred to as the unique solu-
tion) and dependent if it has more than one solution. Two systems of equations
are equivalent if they have the same solution set.
Referring to the three systems in Example 2, the system in part (A) is consistent
and independent with the unique solution x = 4, y = 1. The system in part (B) is
inconsistent. And the system in part (C) is consistent and dependent with an infinite
number of solutions (all points on the two coinciding lines).
Given a system of equations, do not confuse the number of vari-
! CAUTION ables with the number of solutions. The systems of Example 2
involve two variables, x and y. A solution to such a system is a pair of numbers, one
for x and one for y. So the system in Example 2A has two variables, but exactly one
solution, namely, x = 4, y = 1.

Can a consistent and dependent system have exactly two solutions? Exactly three
solutions? Explain.
By graphing a system of two linear equations in two variables, we gain useful

information about the solution set of the system. In general, any two lines in a coor-
dinate plane must intersect in exactly one point, be parallel, or coincide (have identi-
cal graphs). So the systems in Example 2 illustrate the only three possible types of
solutions for systems of two linear equations in two variables. These ideas are sum-
marized in Theorem 1.
THEOREM 1 Possible Solutions to a Linear System

The linear system
ax + by = h
cx + dy = k
must have
(A) Exactly one solution Consistent and independent
or
(B) No solution Inconsistent
or
(C) Infinitely many solutions Consistent and dependent
There are no other possibilities.
In the past, one drawback to solving systems by graphing was the inaccuracy of
hand-drawn graphs. Graphing calculators have changed that. Graphical solutions on a
graphing calculator provide an accurate approximation of the solution to a system of
linear equations in two variables. Example 3 demonstrates this.
EXAMPLE 3 Solving a System Using a Graphing Calculator Solve to two decimal places
using graphical approximation techniques on a graphing calculator:
5x + 2y = 15
2x - 3y = 16
SOLUTION First, solve each equation for y:
5x + 2y = 15 2x - 3y = 16
2y = - 5x + 15 - 3y = - 2x + 16
y = - 2.5x + 7.5 2 16
y = x -
3 3
Next, enter each equation in the graphing calculator (Fig. 2A), graph in an appropri-
ate viewing window, and approximate the intersection point (Fig. 2B).
10
210 10
210
(A) Equation definitions (B) Intersection points

Figure 2
Rounding the values in Figure 2B to two decimal places, we see that the solu-
tion is x = 4.05 and y = - 2.63, or 14.05, - 2.632.
CHECK 5x + 2y = 15 2x - 3y = 16
514.052 + 21 - 2.632 ≟ 15 214.052 - 31 - 2.632 ≟ 16
✓ ✓
14.99 ≈ 15 15.99 ≈ 16
The checks are sufficiently close but, due to rounding, not exact.
Matched Problem 3 Solve to two decimal places using graphical approxima-

tion techniques on a graphing calculator:
2x - 5y = - 25
4x + 3y = 5
Graphical methods help us to visualize a system and its solutions, reveal relation-
ships that might otherwise be hidden, and, with the assistance of a graphing calcula-
tor, provide accurate approximations to solutions.
Substitution
Now we review an algebraic method that is easy to use and provides exact solutions
to a system of two equations in two variables, provided that solutions exist. In this
method, first we choose one of two equations in a system and solve for one variable
in terms of the other. (We make a choice that avoids fractions, if possible.) Then we
substitute the result into the other equation and solve the resulting linear equation in
one variable. Finally, we substitute this result back into the results of the first step to
find the second variable.
EXAMPLE 4 Solving a System by Substitution Solve by substitution:

5x + y = 4
2x - 3y = 5
SOLUTION Solve either equation for one variable in terms of the other; then sub-
stitute into the remaining equation. In this problem, we avoid fractions by choosing
the first equation and solving for y in terms of x:
5x + y = 4 Solve the first equation for y in terms of x.
y = 4 - 5x Substitute into the second equation.
¯˘˙
"
2x - 3y = 5 Second equation
2x - 314 ∙ 5x2 = 5 Solve for x.
2x - 12 + 15x = 5
17x = 17
x = 1
Now, replace x with 1 in y = 4 - 5x to find y:
y = 4 - 5x
y = 4 - 5112
y ∙ -1
The solution is x = 1, y = - 1 or 11, - 12.
CHECK 5x + y = 4 2x - 3y = 5
5112 + 1 - 12 ≟ 4 2112 - 31 - 12 ≟ 5
✓ ✓
4 = 4 5 = 5
Matched Problem 4 Solve by substitution:

3x + 2y = - 2
2x - y = - 6

Return to Example 2 and solve each system by substitution. Based on your results,
describe how you can recognize a dependent system or an inconsistent system when
using substitution.
Elimination by Addition
The methods of graphing and substitution both work well for systems involving two
variables. However, neither is easily extended to larger systems. Now we turn to elimi-
nation by addition. This is probably the most important method of solution. It readily
generalizes to larger systems and forms the basis for computer-based solution methods.
To solve an equation such as 2x - 5 = 3, we perform operations on the equation
until we reach an equivalent equation whose solution is obvious (see Appendix B,
Section B.7).
2x - 5 = 3 Add 5 to both sides.
2x = 8 Divide both sides by 2.
x = 4
Theorem 2 indicates that we can solve systems of linear equations in a similar

manner.
THEOREM 2 Operations That Produce Equivalent Systems

A system of linear equations is transformed into an equivalent system if
(A) Two equations are interchanged.
(B) An equation is multiplied by a nonzero constant.
(C) A constant multiple of one equation is added to another equation.
Any one of the three operations in Theorem 2 can be used to produce an equiva-
lent system, but the operations in parts (B) and (C) will be of most use to us now.
Part (A) becomes useful when we apply the theorem to larger systems. The use of
Theorem 2 is best illustrated by examples.
EXAMPLE 5 Solving a System Using Elimination by Addition Solve the following system
using elimination by addition:
3x - 2y = 8
2x + 5y = - 1
SOLUTION We use Theorem 2 to eliminate one of the variables, obtaining a system

with an obvious solution:
3x - 2y = 8
Multiply the top equation by 5 and the bottom
2x + 5y = - 1 equation by 2 (Theorem 2B).
513x - 2y2 = 5182

212x + 5y2 = 21 - 12
Add the top equation to the bottom equation
15x - 10y = 40
(Theorem 2C), eliminating the y terms.
4x + 10y = - 2 Divide both sides by 19, which is the same as
1
19x = 38 multiplying the equation by 19 (Theorem 2B).
x = 2 This equation paired with either of the
two original equations produces a system
equivalent to the original system.
Knowing that x = 2, we substitute this number back into either of the two original
equations (we choose the second) to solve for y:
2 122 + 5y = - 1
5y = - 5
y = -1
The solution is x = 2, y = - 1 or 12, - 12.
CHECK
3x - 2y = 8 2x + 5y = - 1
3122 - 21 - 12 ≟ 8 2122 + 51 - 12 ≟ - 1
✓ ✓
8 = 8 -1 = -1
Matched Problem 5 Solve the following system using elimination by addition:

5x - 2y = 12
2x + 3y = 1
Let’s see what happens in the elimination process when a system has either no
solution or infinitely many solutions. Consider the following system:
2x + 6y = - 3
x + 3y = 2
Multiplying the second equation by - 2 and adding, we obtain
2x + 6y = - 3
- 2x - 6y = - 4
0 = -7 Not possible
We have obtained a contradiction. The assumption that the original system has solu-
tions must be false. So the system has no solutions, and its solution set is the empty
set. The graphs of the equations are parallel lines, and the system is inconsistent.
Now consider the system
x - 12 y = 4
- 2x + y = - 8
If we multiply the top equation by 2 and add the result to the bottom equation, we
obtain
2x - y = 8
- 2x + y = - 8
0 = 0
Obtaining 0 = 0 implies that the equations are equivalent; that is, their graphs coin-
cide and the system is dependent. If we let x = k, where k is any real number, and
solve either equation for y, we obtain y = 2k - 8. So 1k, 2k - 82 is a solution to
this system for any real number k. The variable k is called a parameter and replac-
ing k with a real number produces a particular solution to the system. For example,
some particular solutions to this system are
k = -1 k = 2 k = 5 k = 9.4
1 - 1, - 102 12, - 42 15, 22 19.4, 10.82
Applications
Many real-world problems are solved readily by constructing a mathematical model
consisting of two linear equations in two variables and applying the solution methods
that we have discussed. We shall examine two applications in detail.
EXAMPLE 6 Diet Jasmine wants to use milk and orange juice to increase the amount of cal-
cium and vitamin A in her daily diet. An ounce of milk contains 37 milligrams
of calcium and 57 micrograms* of vitamin A. An ounce of orange juice contains
5 milligrams of calcium and 65 micrograms of vitamin A. How many ounces of
milk and orange juice should Jasmine drink each day to provide exactly 500 mil-
ligrams of calcium and 1,200 micrograms of vitamin A?
SOLUTION The first step in solving an application problem is to introduce the
proper variables. Often, the question asked in the problem will guide you in this
decision. Reading the last sentence in Example 6, we see that we must determine
a certain number of ounces of milk and orange juice. So we introduce variables to
represent these unknown quantities:
x = number of ounces of milk
y = number of ounces of orange juice
*
A microgram (mg) is one millionth 110 - 6 2 of a gram.
Next, we summarize the given information using a table. It is convenient to organize

the table so that the quantities represented by the variables correspond to columns
in the table (rather than to rows) as shown.
Milk Orange Juice Total Needed

Calcium 37 mg/oz 5 mg/oz 500 mg
Vitamin A 57 μg/oz 65 μg/oz 1,200 μg
Now we use the information in the table to form equations involving x and y:
calcium in x oz calcium in y oz total calcium

a b + a b = a b
of milk of orange juice needed 1mg2
37x + 5y = 500
vitamin A in x oz vitamin A in y oz total vitamin A
a b + a b = a b
of milk of orange juice needed 1mg2
57x + 65y = 1,200
So we have the following model to solve:
37x + 5y = 500
57x + 65y = 1,200
We can multiply the first equation by - 13 and use elimination by addition:
- 481x - 65y = - 6,500 37112.52 + 5y = 500
57x + 65y = 1,200 5y = 37.5
- 424x = - 5,300 y = 7.5
x = 12.5
Drinking 12.5 ounces of milk and 7.5 ounces of orange juice each day will provide
Jasmine with the required amounts of calcium and vitamin A.
100
CHECK 37x + 5y = 500 57x + 65y = 1,200
37112.52 + 517.52 ≟ 500 57112.52 + 6517.52 ≟ 1,200
✓ ✓
500 = 500 1,200 = 1,200
0 25
Figure 3 illustrates a solution to Example 6 using graphical approximation
230
techniques.
Figure 3 Matched Problem 6 Dennis wants to use cottage cheese and yogurt to
y1 = 1500 - 37x2 >5 increase the amount of protein and calcium in his daily diet. An ounce of cot-
y2 = 11,200 - 57x2 >65 tage cheese contains 3 grams of protein and 15 milligrams of calcium. An ounce
of yogurt contains 1 gram of protein and 41 milligrams of calcium. How many
ounces of cottage cheese and yogurt should Dennis eat each day to provide exactly
62 grams of protein and 760 milligrams of calcium?
In a free market economy, the price of a product is determined by the relationship

between supply and demand. Suppliers are more willing to supply a product at higher
prices. So when the price is high, the supply is high. If the relationship between price
and supply is linear, then the graph of the price–supply equation is a line with positive
slope. On the other hand, consumers of a product are generally less willing to buy a
product at higher prices. So when the price is high, demand is low. If the relationship
between price and demand is linear, the graph of the price–demand equation is a line
with negative slope. In a free competitive market, the price of a product tends to move
toward an equilibrium price, in which the supply and demand are equal; that common
value of the supply and demand is the equilibrium quantity. To find the equilibrium
price, we solve the system consisting of the price–supply and price–demand equations.
EXAMPLE 7 Supply and Demand At a price of $1.88 per pound, the supply for cherries in a
large city is 16,000 pounds, and the demand is 10,600 pounds. When the price drops to
$1.46 per pound, the supply decreases to 10,000 pounds, and the demand increases to
12,700 pounds. Assume that the price–supply and price–demand equations are linear.
(A) Find the price–supply equation.
(B) Find the price–demand equation.
(C) Find the supply and demand at a price of $2.09 per pound.
(D) Find the supply and demand at a price of $1.32 per pound.
(E) Use the substitution method to find the equilibrium price and equilibrium demand.
SOLUTION
(A) Let p be the price per pound, and let x be the quantity in thousands of pounds.
Then (16, 1.88) and (10, 1.46) are solutions of the price–supply equation. Use
the point–slope form for the equation of a line, y - y1 = m1x - x1 2, to obtain
the price–supply equation:
1.46 - 1.88
p - 1.88 = 1x - 162 Simplify.
10 - 16
p - 1.88 = 0.071x - 162 Solve for p.
p = 0.07x + 0.76 Price–supply equation
(B) Again, let p be the price per pound, and let x be the quantity in thousands of
pounds. Then (10.6, 1.88) and (12.7, 1.46) are solutions of the price–demand
equation.
1.46 - 1.88
p - 1.88 = 1x - 10.62 Simplify.
12.7 - 10.6
p - 1.88 = - 0.21x - 10.62 Solve for p.
p = - 0.2x + 4 Price–demand equation
(C) Substitute p = 2.09 into the price–supply equation, and also into the price–
demand equation, and solve for x:
Price–supply equation Price–demand equation
p = 0.07x + 0.76 p = - 0.2x + 4
2.09 = 0.07x + 0.76 2.09 = - 0.2x + 4
x = 19 x = 9.55
At a price of $2.09 per pound, the supply is 19,000 pounds of cherries and the
demand is 9,550 pounds. (The supply is greater than the demand, so the price
will tend to come down.)
(D) Substitute p = 1.32 in each equation and solve for x:
Price–supply equation Price–demand equation
p = 0.07x + 0.76 p = - 0.2x + 4
1.32 = 0.07x + 0.76 1.32 = - 0.2x + 4
x = 8 x = 13.4
At a price of $1.32 per pound, the supply is 8,000 pounds of cherries, and the
demand is 13,400 pounds. (The demand is greater than the supply, so the price
will tend to go up.)
(E) We solve the linear system

using substitution (substitute p = - 0.2x + 4 in the first equation):
- 0.2x + 4 = 0.07x + 0.76
- 0.27x = - 3.24 Equilibrium quantity
p x = 12 thousand pounds
4 Now substitute x = 12 into the price–demand equation:

p = - 0.21122 + 4
3
Price per pound ($)
Equilibrium point p = $1.60 per pound Equilibrium price

(12, 1.6)
2 The results are interpreted graphically in Figure 4 (it is customary to refer to
Price–supply
curve
the graphs of price–supply and price–demand equations as “curves” even when
1 Price–demand
they are lines). Note that if the price is above the equilibrium price of $1.60
curve per pound, the supply will exceed the demand and the price will come down.
x
If the price is below the equilibrium price of $1.60 per pound, the demand will
0 5 10 15 20 exceed the supply and the price will go up. So the price will stabilize at $1.60
Thousands of pounds
per pound. At this equilibrium price, suppliers will supply 12,000 pounds of
Figure 4 cherries and consumers will purchase 12,000 pounds.
Matched Problem 7 Find the equilibrium quantity and equilibrium price, and
graph the following price–supply and price–demand equations:
p = 0.08q + 0.66 Price–supply equation
p = - 0.1q + 3 Price–demand equation
Exercises 3.1
Skills Warm-up Exercises A Match each system in Problems 9–12 with one of the following
graphs, and use the graph to solve the system.
W In Problems 1–6, find the x and y coordinates of the intersection y y
of the given lines. (If necessary, review Section A.2).
5 5
1. y = 5x + 7 and the y axis
2. y = 5x + 7 and the x axis
3. 3x + 4y = 72 and the x axis x x
25 5 25 5
4. 3x + 4y = 72 and the y axis

5. 6x - 5y = 120 and x = 5 25 25
6. 6x - 5y = 120 and y = 9 (A) (B)
In Problems 7 and 8, find an equation in point–slope form,

y - y1 = m1x - x1 2, of the line through the given points.
7. (2, 7) and 14, - 52. 8. (5, 39) and 1 - 7, 32.
y y 37. 6x + 0y = 7 38. 6x + 0y = 7
5 5 0x + 9y = - 8 0x + 4y = 9
39. x + y = 0 40. - 2x + y = 0
x - y = 0 5x - y = 0
25 5
x
25 5
x 41. x - 2y = 4 42. x + 6y = - 1
0x + y = 5 0x + y = - 4
43. In a free competitive market, if the supply of a good is greater
25 25 than the demand, will the price tend to go up or come down?
(C) (D) 44. In a free competitive market, if the demand for a good is greater
than the supply, will the price tend to go up or come down?
9. - 4x + 2y = 8 10. x + y = 3
2x - y = 0 2x - y = 0 Problems 45–48 are concerned with the linear system
11. - x + 2y = 5 12. 2x - 4y = - 10 y = mx + b
2x + 3y = - 3 - x + 2y = 5 y = nx + c
Solve Problems 13–16 by graphing. where m, b, n, and c are nonzero constants.
45. If the system has a unique solution, discuss the relationships
13. 3x - y = 2 14. 8x - 3y = 27
among the four constants.
x + 2y = 10 5x + 3y = 12
46. If the system has no solution, discuss the relationships among
15. m + 2n = 4 16. 3u + 5v = 15
the four constants.
2m + 4n = - 8 6u + 10v = - 30
47. If the system has an infinite number of solutions, discuss the
Solve Problems 17–20 using substitution. relationships among the four constants.
17. y = 2x - 3 18. y = x - 4 48. If m = 0, how many solutions does the system have?
x + 2y = 14 x + 3y = 12
In Problems 49–56, use a graphing calculator to find the solution
19. 2x + y = 6 20. 3x - y = 7
to each system. Round any approximate solutions to three decimal
x - y = -3 2x + 3y = 1 places.
Solve Problems 21–24 using elimination by addition. 49. y = 9x - 10 50. y = 3x - 18
21. 3u - 2v = 12 22. 2x - 3y = - 8 y = - 7x + 8 y = - 17x + 51
7u + 2v = 8 5x + 3y = 1 51. y = 0.2x + 0.7 52. y = - 1.7x + 2.3
23. 2m - n = 10 24. 2x + 3y = 1 y = 0.2x - 0.1 y = - 1.7x - 1.3
m - 2n = - 4 3x - y = 7 53. 3x - 2y = 15 54. 3x - 7y = - 20
B Solve Problems 25–34 using substitution or elimination by addition.
4x + 3y = 13 2x + 5y = 8
55. - 2.4x + 3.5y = 0.1 56. 4.2x + 5.4y = - 12.9
25. 6x - 2y = 4 26. 3x + 9y = 6
- 1.7x + 2.6y = - 0.2 6.4x + 3.7y = - 4.5
5x + 3y = 8 4x - 3y = 8
27. 4x - 2y = 10 28. - 5x + 15y = 10 C In Problems 57–62, graph the equations in the same coordinate
system. Find the coordinates of any points where two or more lines
- 6x + 3y = 15 5x - 15y = 10
intersect. Is there a point that is a solution to all three equations?
29. 4x - 2y = 10 30. - 5x + 15y = 10
57. x - 2y = - 6 58. x + y = 3
- 6x + 3y = - 15 5x - 15y = - 10
2x + y = 8 x + 3y = 15
31. 3m + 5n = 7 32. 8m - 9n = 12
x + 2y = - 2 3x - y = 5
2m + 10n = 12 3m - 15n = 51
59. x + y = 1 60. x - y = 6
33. x + y = 1 34. x + 2y = 3
x - 2y = - 8 x - 2y = 8
0.3x + 0.5y = 0.7 0.2x + 0.5y = 0.4
3x + y = - 3 x + 4y = - 4
In Problems 35–42, solve the system. Note that each solution can 61. 4x - 3y = - 24 62. 2x + 3y = 18
be found mentally, without the use of a calculator or pencil-and-
paper calculation; try to visualize the graphs of both lines. 2x + 3y = 12 2x - 6y = - 6
8x - 6y = 24 4x + 6y = - 24
35. x + 0y = 7 36. x + 0y = - 4
63. The coefficients of the three systems given below are similar.
0x + y = 3 0x + y = 9 One might guess that the solution sets to the three systems
would be nearly identical. Develop evidence for or against
this guess by considering graphs of the systems and solutions the price increases to $5.10 per bushel, the annual supply in-
obtained using substitution or elimination by addition. creases to 2.1 billion bushels, and the annual demand decreases
(A) 5x + 4y = 4 (B) 5x + 4y = 4 to 1.8 billion bushels. Assume that the price–supply and price–
demand equations are linear. (Source: U.S. Census Bureau)
11x + 9y = 4 11x + 8y = 4
(C) 5x + 4y = 4 (A) Find the price–supply equation.
10x + 8y = 4 (B) Find the price–demand equation.
64. Repeat Problem 63 for the following systems: (C) Find the equilibrium price and quantity.
(A) 6x - 5y = 10 (B) 6x - 5y = 10
(D) Graph the two equations in the same coordinate system
- 13x + 11y = - 20 - 13x + 10y = - 20 and identify the equilibrium point, supply curve, and
(C) 6x - 5y = 10 demand curve.
- 12x + 10y = - 20
68. Supply and demand for corn. At $2.13 per bushel, the
annual supply for corn in the Midwest is 8.9 billion bushels
and the annual demand is 6.5 billion bushels. When the price
Applications
falls to $1.50 per bushel, the annual supply decreases to 8.2
billion bushels and the annual demand increases to 7.4 billion
bushels. Assume that the price–supply and price–demand
65. Supply and demand for T-shirts. Suppose that the supply equations are linear. (Source: U.S. Census Bureau)
and demand equations for printed T-shirts for a particular
week are (A) Find the price–supply equation.
p = 0.7q + 3 Price–supply equation (B) Find the price–demand equation.

p = - 1.7q + 15 Price–demand equation (C) Find the equilibrium price and quantity.
where p is the price in dollars and q is the quantity in hundreds. (D) Graph the two equations in the same coordinate system
(A) Find the supply and demand (to the nearest unit) if and identify the equilibrium point, supply curve, and
T-shirts are $4 each. Discuss the stability of the T-shirt demand curve.
market at this price level. 69. Break-even analysis. A small plant manufactures riding
(B) Find the supply and demand (to the nearest unit) if lawn mowers. The plant has fixed costs (leases, insurance,
T-shirts are $9 each. Discuss the stability of the T-shirt etc.) of $48,000 per day and variable costs (labor, materials,
market at this price level. etc.) of $1,400 per unit produced. The mowers are sold for
$1,800 each. So the cost and revenue equations are
(C) Find the equilibrium price and quantity.
y = 48,000 + 1,400x Cost equation
(D) Graph the two equations in the same coordinate system
y = 1,800x Revenue equation
and identify the equilibrium point, supply curve, and
demand curve. where x is the total number of mowers produced and sold
66. Supply and demand for caps. Suppose that the supply each day. The daily costs and revenue are in dollars.
and demand for printed caps for a particular week are (A) How many units must be manufactured and sold each day
for the company to break even?
p = 0.4q + 3.2 Price–supply equation
p = - 1.9q + 17 Price–demand equation (B) Graph both equations in the same coordinate system and
show the break-even point. Interpret the regions between
where p is the price in dollars and q is the quantity in
the lines to the left and to the right of the break-even
hundreds.
point.
(A) Find the supply and demand (to the nearest unit) if caps
are $4 each. Discuss the stability of this particular cap 70. Break-even analysis. Repeat Problem 69 with the cost and
market at this price level. revenue equations
(B) Find the supply and demand (to the nearest unit) if caps y = 65,000 + 1,100x Cost equation
are $9 each. Discuss the stability of the cap market at y = 1,600x Revenue equation
this price level.
(C) Find the equilibrium price and quantity. 71. Break-even analysis. A company markets exercise DVDs
that sell for $19.95, including shipping and handling. The
(D) Graph the two equations in the same coordinate system monthly fixed costs (advertising, rent, etc.) are $24,000 and the
and identify the equilibrium point, supply curve, and variable costs (materials, shipping, etc) are $7.45 per DVD.
demand curve.
(A) Find the cost equation and the revenue equation.
67. Supply and demand for soybeans. At $4.80 per bushel,
the annual supply for soybeans in the Midwest is 1.9 billion (B) How many DVDs must be sold each month for the
bushels, and the annual demand is 2.0 billion bushels. When company to break even?
(C) Graph the cost and revenue equations in the same coor- acid. How many bags of each brand should be used to pro-
dinate system and show the break-even point. Interpret vide the required amounts of nitrogen and phosphoric acid?
the regions between the lines to the left and to the right
79. Electronics. A supplier for the electronics industry manufac-
of the break-even point.
tures keyboards and screens for graphing calculators at plants
72. Break-even analysis. Repeat Problem 71 if the monthly fixed in Mexico and Taiwan. The hourly production rates at each
costs increase to $27,200, the variable costs increase to $9.15, plant are given in the table. How many hours should each
and the company raises the selling price of the DVDs to $21.95. plant be operated to exactly fill an order for 4,000 keyboards
and 4,000 screens?
73. Delivery charges. Global Parcel, an international package
delivery service, charges a base price for overnight delivery Plant Keyboards Screens
of packages weighing 1 kg or less and a surcharge for each Mexico 40 32
additional kg (or fraction thereof). A customer is billed
Taiwan 20 32
€13.68 for shipping a 3-kg package and €24.60 for a 10-kg
package. Find the base price and the surcharge for each ad- 80. Sausage. A company produces Italian sausages and brat-
ditional kg. wursts at plants in Green Bay and Sheboygan. The hourly
74. Delivery charges. Refer to Problem 73. Parcel Express, a production rates at each plant are given in the table. How
competing overnight delivery service, informs the customer many hours should each plant operate to exactly fill an order
in Problem 73 that they would ship the 3-kg package for for 62,250 Italian sausages and 76,500 bratwursts?
€14.75 and the 10-kg package for €23.78.
Plant Italian Sausage Bratwurst
(A) If Parcel Express computes its cost in the same manner Green Bay 800 800
as Global Parcel, find the base price and the surcharge Sheboygan 500 1,000
for Parcel Express.
81. Physics. An object dropped off the top of a tall building
(B) Devise a simple rule that the customer can use to
falls vertically with constant acceleration. If s is the distance
choose the cheaper of the two services for each package
of the object above the ground (in feet) t seconds after its
shipped. Justify your answer.
release, then s and t are related by an equation of the form
75. Coffee blends. A coffee company uses Colombian and s = a + bt 2, where a and b are constants. Suppose the object
Brazilian coffee beans to produce two blends, robust and mild. is 180 feet above the ground 1 second after its release and
A pound of the robust blend requires 12 ounces of Colombian 132 feet above the ground 2 seconds after its release.
beans and 4 ounces of Brazilian beans. A pound of the mild
(A) Find the constants a and b.
blend requires 6 ounces of Colombian beans and 10 ounces of
Brazilian beans. Coffee is shipped in 132-pound burlap bags. (B) How tall is the building?
The company has 50 bags of Colombian beans and 40 bags
(C) How long does the object fall?
of Brazilian beans on hand. How many pounds of each blend
should the company produce in order to use all the available 82. Physics. Repeat Problem 81 if the object is 240 feet above
beans? the ground after 1 second and 192 feet above the ground after
2 seconds.
76. Coffee blends. Refer to Problem 75.
83. Earthquakes. An earthquake emits a primary wave and a
(A) If the company decides to discontinue production of the
secondary wave. Near the surface of the Earth the primary
robust blend and produce only the mild blend, how many
wave travels at 5 miles per second and the secondary wave at
pounds of the mild blend can the company produce?
3 miles per second. From the time lag between the two waves
How many beans of each type will the company use?
arriving at a given receiving station, it is possible to estimate
Are there any beans that are not used?
the distance to the quake. Suppose a station measured a
(B) Repeat part (A) if the company decides to discontinue time difference of 16 seconds between the arrival of the two
production of the mild blend and produce only the robust waves. How long did each wave travel, and how far was the
blend. earthquake from the station?
77. Animal diet. Animals in an experiment are to be kept under 84. Sound waves. A ship using sound-sensing devices above and
a strict diet. Each animal should receive 20 grams of protein below water recorded a surface explosion 6 seconds sooner
and 6 grams of fat. The laboratory technician is able to by its underwater device than its above-water device. Sound
purchase two food mixes: Mix A has 10% protein and 6% fat; travels in air at 1,100 feet per second and in seawater at 5,000
mix B has 20% protein and 2% fat. How many grams of each feet per second. How long did it take each sound wave to reach
mix should be used to obtain the right diet for one animal? the ship? How far was the explosion from the ship?
78. Fertilizer. A fruit grower uses two types of fertilizer in an 85. Psychology. People approach certain situations with “mixed
orange grove, brand A and brand B. Each bag of brand A con- emotions.” For example, public speaking often brings forth the
tains 8 pounds of nitrogen and 4 pounds of phosphoric acid. positive response of recognition and the negative response of
Each bag of brand B contains 7 pounds of nitrogen and failure. Which dominates? J. S. Brown, in an experiment on
6 pounds of phosphoric acid. Tests indicate that the grove approach and avoidance, trained rats by feeding them from a
needs 720 pounds of nitrogen and 500 pounds of phosphoric goal box. The rats received mild electric shocks from the same
goal box. This established an approach–avoidance conflict 2. (A) x = 2, y = 2

relative to the goal box. Using an appropriate apparatus, (B) Infinitely many solutions
Brown arrived at the following relationships:
(C) No solution
p = - 15 d + 70 Approach equation 3. x = - 1.92, y = 4.23
4. x = - 2, y = 2
p = - 43 d + 230 Avoidance equation
5. x = 2, y = - 1
where 30 … d … 172.5. The approach equation gives the
6. 16.5 oz of cottage cheese, 12.5 oz of yogurt
pull (in grams) toward the food goal box when the rat is
placed d centimeters away from it. The avoidance equation 7. Equilibrium quantity = 13 thousand pounds; equilibrium
gives the pull (in grams) away from the shock goal box when price = $1.70 per pound
the rat is placed d centimeters from it.
p
(A) Graph the approach equation and the avoidance equation
in the same coordinate system. 3
(B) Find the value of d for the point of intersection of these Equilibrium point
(13, 1.7)
Price per pound ($)

two equations.
2 Supply curve
(C) What do you think the rat would do when placed the dis-
tance d from the box found in part (B)? Demand curve
(Source: Journal of Comparative and Physiological Psychology, 1
41:450–465.)
q
Answers to Matched Problems 0 5 10 15 20
1. x = - 2, y = - 1 y Thousands of pounds
2x - y = -3 5
21 - 22 - 1 - 12 ≟ - 3
✓ (22, 21)
-3 = -3
x
x + 2y = - 4 25 5
1 - 22 + 21 - 12 ≟ - 4
✓
-4 = -4 25
2x 2 y 5 23 x 1 2y 5 24
3.2 Systems of Linear Equations and Augmented Matrices

■■ Matrices Most linear systems of any consequence involve large numbers of equations and
■■ Solving Linear Systems Using
variables. It is impractical to try to solve such systems by hand. In the past, these
Augmented Matrices complex systems could be solved only on large computers. Now there are a wide
array of approaches to solving linear systems, ranging from graphing calculators
■■ Summary to software and spreadsheets. In the rest of this chapter, we develop several matrix
methods for solving systems with the understanding that these methods are gener-
ally used with a graphing calculator. It is important to keep in mind that we are
not presenting these techniques as efficient methods for solving linear systems by
hand. Instead, we emphasize formulation of mathematical models and interpreta-
tion of the results—two activities that graphing calculators cannot perform for you.
Matrices
In solving systems of equations using elimination by addition, the coefficients of the
variables and the constant terms played a central role. The process can be made more
efficient for generalization and computer work by the introduction of a mathematical
SECTION 3.2 Systems of Linear Equations and Augmented Matrices 167
form called a matrix. A matrix is a rectangular array of numbers written within

brackets. Two examples are
-4 5 12
1 -4 5 0 1 8
A = c d B = D T (1)
7 0 -2 -3 10 9
-6 0 -1
Each number in a matrix is called an element of the matrix. Matrix A has 6 elements
arranged in 2 rows and 3 columns. Matrix B has 12 elements arranged in 4 rows and
3 columns. If a matrix has m rows and n columns, it is called an m : n matrix (read
“m by n matrix”). The expression m * n is called the size of the matrix, and the num-
bers m and n are called the dimensions of the matrix. It is important to note that the
number of rows is always given first. Referring to equations (1), A is a 2 * 3 matrix
and B is a 4 * 3 matrix. A matrix with n rows and n columns is called a square
matrix of order n. A matrix with only 1 column is called a column matrix, and a
matrix with only 1 row is called a row matrix.
3 * 3 4 * 1 1 * 4
3
0.5 0.2 1.0
-2 1 2
C 0.0 0.3 0.5 S D T c2 0 - d
1 2 3
0.7 0.0 0.2
0
Square matrix of order 3 Column matrix Row matrix
The position of an element in a matrix is given by the row and column contain-
ing the element. This is usually denoted using double subscript notation aij, where i
is the row and j is the column containing the element aij, as illustrated below:
1 -4 5 a11 = 1, a12 = - 4, a13 = 5

A = c d
7 0 -2 a21 = 7, a22 = 0, a23 = - 2
Note that a12 is read “a sub one two” (not “a sub twelve”). The elements a11 = 1 and
a22 = 0 make up the principal diagonal of A. In general, the principal diagonal of a
matrix A consists of the elements a11, a22, a33, c.
Remark—Most graphing calculators are capable of storing and manipulating matri-

ces. Figure 1 shows matrix A displayed in the editing screen of a graphing calculator.
The size of the matrix is given at the top of the screen. The position and value of the
currently selected element is given at the bottom. Note that a comma is used in the
notation for the position. This is common practice on many graphing calculators but
not in mathematical literature. In a spreadsheet, matrices are referred to by their loca-
tion (upper left corner to lower right corner), using either row and column numbers
(Fig. 2A) or row numbers and column letters (Fig. 2B).
Figure 1 Matrix notation on a
graphing calculator
(A) Location of matrix A: (B) Location of matrix A:

R1C1:R2C3 D5:F6
Figure 2 Matrix notation in a spreadsheet
Matrices serve as a shorthand for solving systems of linear equations. Associated

with the system
2x - 3y = 5
(2)
x + 2y = - 3
are its coefficient matrix, constant matrix, and augmented matrix:
Coefficient Constant Augmented
matrix matrix matrix
2 -3 5 2 -3 5
c d c d c ` d
1 2 -3 1 2 -3
Note that the augmented matrix is just the coefficient matrix, augmented by the
constant matrix. The vertical bar is included only as a visual aid to separate the coef-
ficients from the constant terms. The augmented matrix contains all of the essential
information about the linear system—everything but the names of the variables.
For ease of generalization to the larger systems in later sections, we will change
the notation for the variables in system (2) to a subscript form. That is, in place of x
and y, we use x1 and x2, respectively, and system (2) is rewritten as
2x1 - 3x2 = 5
x1 + 2x2 = - 3
In general, associated with each linear system of the form
a11x1 + a12x2 = k1
(3)
a21x1 + a22x2 = k2
where x1 and x2 are variables, is the augmented matrix of the system:
Column 1 (C1)
Column 2 (C2)
Column 3 (C3)
"
"
"
a a12 k1 d Row 1 1R1 2

c 11 ` d
a21 a22 k2 d Row 2 1R2 2
This matrix contains the essential parts of system (3). Our objective is to learn how to
manipulate augmented matrices in order to solve system (3), if a solution exists. The ma-
nipulative process is closely related to the elimination process discussed in Section 3.1.
Recall that two linear systems are said to be equivalent if they have the same
solution set. In Theorem 2, Section 3.1, we used the operations listed below to trans-
form linear systems into equivalent systems:
(A) Two equations are interchanged.
(B) An equation is multiplied by a nonzero constant.
(C) A constant multiple of one equation is added to another equation.
Paralleling the earlier discussion, we say that two augmented matrices are row
equivalent, denoted by the symbol ∙ placed between the two matrices, if they are
augmented matrices of equivalent systems of equations. How do we transform aug-
mented matrices into row-equivalent matrices? We use Theorem 1, which is a direct
consequence of the operations listed in Section 3.1.
THEOREM 1 Operations That Produce Row-Equivalent Matrices

An augmented matrix is transformed into a row-equivalent matrix by performing
any of the following row operations:
(A) Two rows are interchanged 1Ri 4 Rj 2.
(B) A row is multiplied by a nonzero constant 1kRi S Ri 2.
(C) A constant multiple of one row is added to another row 1kRj + Ri S Ri 2.
Note: The arrow S means “replaces.”
Solving Linear Systems Using Augmented Matrices

We illustrate the use of Theorem 1 by several examples.
EXAMPLE 1 Solving a System Using Augmented Matrix Methods Solve using augmented
matrix methods:
3x1 + 4x2 = 1
(4)
x1 - 2x2 = 7
SOLUTION We start by writing the augmented matrix corresponding to system (4):
3 4 1
c ` d (5)
1 -2 7
Our objective is to use row operations from Theorem 1 to try to transform matrix
(5) into the form
1 0 m
c ` d (6)
0 1 n
where m and n are real numbers. Then the solution to system (4) will be obvious,
since matrix (6) will be the augmented matrix of the following system (a row in an
augmented matrix always corresponds to an equation in a linear system):
x1 = m x1 + 0x2 = m
x2 = n 0x1 + x2 = n
Now we use row operations to transform matrix (5) into form (6).
Step 1 To get a 1 in the upper left corner, we interchange R1 and R2 (Theorem 1A):
3 4 1 R1 4 R2 1 -2 7
c ` d ∙ c ` d
1 -2 7 3 4 1
Step 2 To get a 0 in the lower left corner, we multiply R1 by 1 - 32 and add to R2
(Theorem 1C)—this changes R2 but not R1. Some people find it useful to write
1 - 3R1 2 outside the matrix to help reduce errors in arithmetic, as shown:
1 -2 1 1 -2 7
c ` d 1 - 32R ∼
+ R2 S R2 c ` d
3 4 7 ¯˘˙1 0 10 - 20
-3 6 - 21 "
1
Step 3 To get a 1 in the second row, second column, we multiply R2 by 10
(Theorem 1B):
1 -2 7 1 ∼ R2 c 1 -2 7
c ` d 10 R2 S ` d
0 10 - 20 0 1 -2
Step 4 To get a 0 in the first row, second column, we multiply R2 by 2 and add the
result to R1 (Theorem 1C)—this changes R1 but not R2:
0 2 -4 "
1 -2 7 ¸˝˛ 1 0 3
c ` d 2R2 + R1 S R1 c ` d
0 1 -2 ∼ 0 1 -2
We have accomplished our objective! The last matrix is the augmented matrix
for the system
x1 = 3 x1 + 0x2 = 3
(7)
x2 = - 2 0x1 + x2 = - 2
Since system (7) is equivalent to system (4), our starting system, we have solved
system (4); that is, x1 = 3 and x2 = - 2.
CHECK 3x1 + 4x2 = 1 x1 - 2x2 = 7

3132 + 41 - 22 ≟ 1 3 - 21 - 22 ≟ 7
✓ ✓
1 = 1 7 = 7
The preceding process may be written more compactly as follows:
⁄
Step 1: 3 4 1 R1 4 R2
c ` d
Need a 1 here. 1 -2 7
1 -2 7 - 32R1 + R3 S R2
1¯˘˙
Step 2: ∼c ` d
Need a 0 here. ⁄3 4 1
-3 6 - 21
⁄
Step 3: 1 -2 7
∼c ` d 1
S R2
Need a 1 here. 0 10 - 20 10 R2
⁄
0 2 -4
⁄
1 -2 7 ¯ ˘˙
∼c ` d 2R2 + R1 S R1
⁄
Step 4: 0 1 -2
Need a 0 here.
1 0 3
∼c ` d
0 1 -2
Therefore, x1 = 3 and x2 = - 2.
Matched Problem 1 Solve using augmented matrix methods:

2x1 - x2 = - 7
x1 + 2x2 = 4
Many graphing calculators can perform row operations. Figure 3 shows the re-
sults of performing the row operations used in the solution of Example 1. Consult
your manual for the details of performing row operations on your graphing calculator.
(B) (23)R1 1 R2
(A) R1
QW R2 W R2
1
(C) 22R
10 2 W R2 (D) 2R2 1 R1
W R1
Figure 3 Row operations on a graphing calculator

The summary following the solution of Example 1 shows five augmented matrices.
Write the linear system that each matrix represents, solve each system graphically,
and discuss the relationships among these solutions.
matrix methods:
2x1 - 3x2 = 6
1
3x1 + 4x2 =
2
SOLUTION
"
1
Step 1: 2 -3 6 2 R1 S R1
c ` 1d
Need a 1 here. 3 4 2
1 - 32 3
Step 2: ∼c ` 1d 1 - 32R1 + R2 S R2
"3 4 2
Need a 0 here. ¯˘˙
-3 9
2 -9"
1 - 32 3
∼c " 17
` d
Step 3: 0 2 - 17
2
2
17 R2 S R2
Need a 1 here.
0 3
2 - 32 "
1 -" 32 3 ¯˘˙
3
∼c ` d 2 R2 + R1 S R1
Step 4: 0 1 -1
Need a 0 here.
3
1 0 2
∼c ` d
0 1 -1
3
So, x1 = 2 and x2 = - 1. The check is left for you.

5x1 - 2x2 = 11
5
2x1 + 3x2 =
2
matrix methods:
2x1 - x2 = 4
(8)
- 6x1 + 3x2 = - 12
SOLUTION
1
2 -1 4 2 R1 S R1 (to get a 1 in the upper left corner)
c ` d 1
S R2 (this simplifies R2)
-6 3 - 12 3 R2
1 - 12 2
∼c ` d 2R + R2 S R2 (to get a 0 in the lower left corner)
1
-2 1 -4 ˙˘¯
2 -1 4 "
1 - 12 2
∼c ` d
0 0 0
The last matrix corresponds to the system
1 1
x1 - x = 2 x1 - x = 2
2 2 2 2
(9)
0 = 0 0x1 + 0x2 = 0
This system is equivalent to the original system. Geometrically, the graphs of the
two original equations coincide, and there are infinitely many solutions. In general,
if we end up with a row of zeros in an augmented matrix for a two-equation, two-
variable system, the system is dependent, and there are infinitely many solutions.
We represent the infinitely many solutions using the same method that was used
in Section 3.1, that is, by introducing a parameter. We start by solving x1 - 12 x2 = 2,
the first equation in system (9), for either variable in terms of the other. We choose
to solve for x1 in terms of x2 because it is easier:
1
x1 = x2 + 2 (10)
2
Now we introduce a parameter t (we can use other letters, such as k, s, p, q, and so
on, to represent a parameter also). If we let x2 = t, then for any real number t,
1
x1 = t + 2
2
(11)
x2 = t
represents a solution of system (8). Using ordered-pair notation, we write: For any
real number t,
1
a t + 2, tb (12)
2
is a solution of system (8). More formally, we write
1
solution set = e a t + 2, tb ` t ∊ R f (13)
2
Typically we use the less formal notations (11) or (12) to represent the solution set
for problems of this type.
CHECK The following is a check that system (11) provides a solution to system (8)
for any real number t:
2x1 - x2 = 4 - 6x1 + 3x2 = - 12
1 1
2a t + 2b - t ≟ 4 - 6 a t + 2 b + 3t ≟ - 12
2 2
t + 4 - t ≟4 - 3t - 12 + 3t ≟ - 12
✓ ✓
4 = 4 - 12 = - 12
- 2x1 + 6x2 = 6
3x1 - 9x2 = - 9

The solution of Example 3 involved three augmented matrices. Write the linear
system that each matrix represents, solve each system graphically, and discuss the
relationships among these solutions.
matrix methods:
2x1 + 6x2 = - 3
x1 + 3x2 = 2
SOLUTION 2 6 -3
c 2 d R1 4 R2
1 3 2
1 3 2
2
∼c d 1 - 22R1 + R2 S R2
2 6 -3 ˙˘¯
-2 -6 -4 "
1 3 2 2
∼c d R2 implies the contradiction 0 = - 7.
0 0 -7
This is the augmented matrix of the system
x1 + 3x2 = 2 x1 + 3x2 = 2
0 = -7 0x1 + 0x2 = - 7
The second equation is not satisfied by any ordered pair of real numbers. As we saw
in Section 3.1, the original system is inconsistent and has no solution. If in a row of
an augmented matrix, we obtain all zeros to the left of the vertical bar and a nonzero
number to the right, the system is inconsistent and there are no solutions.

2x1 - x2 = 3
4x1 - 2x2 = - 1
Summary
Examples 2, 3, and 4 illustrate the three possible solution types for a system of two
linear equations in two variables, as discussed in Theorem 1, Section 3.1. Examining
the final matrix form in each of these solutions leads to the following summary.
SUMMARY Possible Final Matrix Forms for a System of Two Linear

Form 1: Exactly one solution Form 2: Infinitely many solutions Form 3: No solution
(consistent and independent) (consistent and dependent) (inconsistent)
1 0 2 m 1 m 2 n 1 m 2 n
c d c d c d
0 1 n 0 0 0 0 0 p
m, n, p are real numbers; p ∙ 0
The process of solving systems of equations described in this section is referred

to as Gauss–Jordan elimination. We formalize this method in the next section so
that it will apply to systems of any size, including systems where the number of equa-
tions and the number of variables are not the same.
Exercises 3.2
W Skills Warm-up Exercises 4 0 - 10 1 -3 5
21. c ` d 22. c ` d
Problems 1–14 refer to the following matrices: (If necessary, 0 8 40 0 1 7
review the terminology at the beginning of Section 3.2.)
Perform the row operations indicated in Problems 23–34 on the
2 -4 0 following matrix:
A = c d
6 1 -5
2 -4 6
-1 9 0 c ` d
1 -3 5
B = C -4 8 7S
2 4 0 23. R1 4 R2 24. R2 4 R1
25. 2R2 S R2 26. - 2R2 S R2
C = 32 - 3 04
27. R1 + R2 S R1 28. R1 + R2 S R2
-5 29. - 12 R1 S R1 30. 1
2 R1 S R1
D = c d
8
31. 1 - 12R2 + R1 S R1 32. 1 - 22R2 + R1 S R1
33. 1 - 32R2 + R1 S R1 34. 1 - 12 2R1 + R2 S R2
1. How many elements are there in A? In C? Each of the matrices in Problems 35–42 is the result of perform-
ing a single row operation on the matrix A shown below. Identify
2. How many elements are there in B? In D?
the row operation.
3. What is the size of B? Of D?
-1 2 2 -3
4. What is the size of A? Of C? A = c d
6 -3 12
5. Which of the matrices is a column matrix?
-1 2 -3 -6 12 - 18
6. Which of the matrices is a row matrix? 35. c ` d 36. c ` d
2 -1 4 6 -3 12
7. Which of the matrices is a square matrix? -1 2 -3 3 0 5
37. c ` d 38. c ` d
8. Which of the matrices does not contain the element 0? 0 9 -6 6 -3 12
A 9. List the elements on the principal diagonal of A. 1 1 1 -1 2 -3

39. c ` d 40. c ` d
6 -3 12 2 5 0
10. List the elements on the principal diagonal of B.
6 -3 12 -1 2 -3
11. For matrix B, list the elements b21, b12, b33. 41. c ` d 42. c ` d
-1 2 -3 0 9 -6
12. For matrix A, list the elements a21, a12.
Solve Problems 43–46 using augmented matrix methods. Graph
13. For matrix C, find c11 + c12 + c13.
each solution set. Discuss the differences between the graph of an
14. For matrix D, find d11 + d21. equation in the system and the graph of the system’s solution set.
43. 3x1 - 2x2 = 6 44. x1 - 2x2 = 5
In Problems 15–18, write the coefficient matrix and the augmented 4x1 - 3x2 = 6 - 2x1 + 4x2 = - 10
matrix of the given system of linear equations.
45. 3x1 - 2x2 = - 3 46. x1 - 2x2 = 1
15. 3x1 + 5x2 = 8 16. 5x1 + 7x2 = 11 - 6x1 + 4x2 = 6 - 2x1 + 5x2 = 2
2x1 - 4x2 = - 7 - 4x1 + 9x2 = 16
Solve Problems 47 and 48 using augmented matrix methods.
17. x1 + 4x2 = 15 18. 5x1 - x2 = 10 Write the linear system represented by each augmented
6x1 = 18 3x2 = 21 matrix in your solution, and solve each of these systems
graphically. Discuss the relationships among the solutions
of these systems.
In Problems 19–22, write the system of linear equations that is
represented by the given augmented matrix. Assume that the vari- 47. x1 + x2 = 5 48. x1 - x2 = 2
ables are x1 and x2. x1 - x2 = 1 x1 + x2 = 6
2 5 7 0 3 15
19. c ` d 20. c ` d
1 4 9 -8 2 25
SECTION 3.3 Gauss–Jordan Elimination 175
Each of the matrices in Problems 49–54 is the final matrix form 73. - 6x1 + 8x2 = - 18 74. - 2x1 + 4x2 = 4
for a system of two linear equations in the variables x1 and x2. 9x1 - 12x2 = 27 3x1 - 6x2 = - 6
Write the solution of the system.
C Solve Problems 75–80 using augmented matrix methods.
49. c
1 0 2 -4 d 50. c
1 0 2 -4 d 75. 3x1 - x2 = 7 76. 2x1 - 3x2 = - 8
0 1 6 0 1 7
2x1 + 3x2 = 1 5x1 + 3x2 = 1
1 3 2 1 -2 7 77. 3x1 + 2x2 = 4 78. 5x1 + 2x2 = 23
51. c 2 d 52. c 2 d
0 0 4 0 0 -9 2x1 - x2 = 5 7x1 - 6x2 = - 3
79. 0.2x1 - 0.5x2 = 0.07 80. 0.3x1 - 0.6x2 = 0.18
-2
2 -9 d
1 2 15 1 7
53. c d 54. c 0.8x1 - 0.3x2 = 0.79 0.5x1 - 0.2x2 = 0.54
0 0 0 0 0 0
Solve Problems 81–84 using augmented matrix methods. Use a
B Solve Problems 55–74 using augmented matrix methods. graphing calculator to perform the row operations.
55. x1 - 2x2 = 1 56. x1 + 3x2 = 1
81. 0.8x1 + 2.88x2 = 4
2x1 - x2 = 5 3x1 - 2x2 = 14
1.25x1 + 4.34x2 = 5
57. x1 - 3x2 = 9 58. x1 - 3x2 = - 5
- 2x1 + x2 = - 8 - 3x1 - x2 = 5 82. 2.7x1 - 15.12x2 = 27
59. 3x1 - x2 = 2 60. 2x1 + x2 = 0 3.25x1 - 18.52x2 = 33
x1 + 2x2 = 10 x1 - 2x2 = - 5 83. 4.8x1 - 40.32x2 = 295.2
61. x1 + 2x2 = 4 62. 2x1 - 3x2 = - 2 - 3.75x1 + 28.7x2 = - 211.2
2x1 + 4x2 = - 8 - 4x1 + 6x2 = 7 84. 5.7x1 - 8.55x2 = - 35.91
63. 2x1 + x2 = 6 64. 3x1 - x2 = - 5 4.5x1 + 5.73x2 = 76.17
x1 - x2 = - 3 x1 + 3x2 = 5
65. 3x1 - 6x2 = - 9 66. 2x1 - 4x2 = - 2 Answers to Matched Problems
- 2x1 + 4x2 = 6 - 3x1 + 6x2 = 3 1. x1 = - 2, x2 = 3
67. 4x1 - 2x2 = 2 68. - 6x1 + 2x2 = 4 2. x1 = 2, x2 = - 12
- 6x1 + 3x2 = - 3 3x1 - x2 = - 2
3. The system is dependent. For t any real number, a solution is
69. 2x1 + x2 = 1 70. 2x1 - x2 = - 8 x1 = 3t - 3, x2 = t.
4x1 - x2 = - 7 2x1 + x2 = 8 4. Inconsistent—no solution
71. 4x1 - 6x2 = 8 72. 2x1 - 4x2 = - 4
- 6x1 + 9x2 = - 10 - 3x1 + 6x2 = 4

■■ Reduced Matrices Now that you have had some experience with row operations on simple augmented
■■ Solving Systems by Gauss–Jordan
matrices, we consider systems involving more than two variables. We will not require
Elimination a system to have the same number of equations as variables. Just as for systems of
two linear equations in two variables, any linear system, regardless of the number of
■■ Application equations or number of variables, has either
1. Exactly one solution (consistent and independent), or
2. Infinitely many solutions (consistent and dependent), or
3. No solution (inconsistent).
Reduced Matrices
In the preceding section we used row operations to transform the augmented matrix
for a system of two equations in two variables,
a11 a12 k1 a11x1 + a12x2 = k1

c ` d
a21 a22 k2 a21x1 + a22x2 = k2
into one of the following simplified forms:

Form 1 Form 2 Form 3
1 0 m 1 m n 1 m n
c ` d c ` d c ` d (1)
0 1 n 0 0 0 0 0 p
where m, n, and p are real numbers, p ∙ 0. Each of these reduced forms represents
a system that has a different type of solution set, and no two of these forms are row
equivalent.
For large linear systems, it is not practical to list all such simplified forms; there are
too many of them. Instead, we give a general definition of a simplified form called a
reduced matrix, which can be applied to all matrices and systems, regardless of size.
DEFINITION Reduced Form

A matrix is said to be in reduced row echelon form, or, more simply, in reduced
form, if
1. Each row consisting entirely of zeros is below any row having at least one
nonzero element.
2. The leftmost nonzero element in each row is 1.
3. All other elements in the column containing the leftmost 1 of a given row are
zeros.
4. The leftmost 1 in any row is to the right of the leftmost 1 in the row above.
The following matrices are in reduced form. Check each one carefully to con-
vince yourself that the conditions in the definition are met.
Note that a row of a reduced matrix may have a leftmost 1 in the last column.
1 0 0 2 1 0 3
1 0 2
c 2 d C0 1 0 3 -1 S C0 1 3 -1 S
0 1 -3
0 0 1 3 0 0 0
1 4 0 0 -3 1 0 4 0
C0 0 1 0 3 2S C0 1 3 3 0S
0 0 0 1 6 0 0 0 1
EXAMPLE 1 Reduced Forms The following matrices are not in reduced form. Indicate which
condition in the definition is violated for each matrix. State the row operation(s)
required to transform the matrix into reduced form, and find the reduced form.
0 1 2 -2 1 2 -2 2 3
(A) c d (B) c d
1 0 3 0 0 1 -1
1 0 -3 1 0 0 -1
(C) C 0 0 3 0S (D) C 0 2 0 3 3S
0 1 -2 0 0 1 -5
SOLUTION
(A) Condition 4 is violated: The leftmost 1 in row 2 is not to the right of the left-
most 1 in row 1. Perform the row operation R1 4 R2 to obtain
1 0 2 3
c d
0 1 -2
(B) Condition 3 is violated: The column containing the leftmost 1 in row 2 has a
nonzero element above the 1. Perform the row operation 2R2 + R1 S R1 to
obtain
1 2 0 2 1
c d
0 0 1 -1
(C) Condition 1 is violated: The second row contains all zeros and is not below any
row having at least one nonzero element. Perform the row operation R2 4 R3
to obtain
1 0 -3
C0 1 3 -2 S
0 0 0
(D) Condition 2 is violated: The leftmost nonzero element in row 2 is not a 1. Per-
form the row operation 12R2 S R2 to obtain
1 0 0 -1
C0 1 0 3 3
S
2
0 0 1 -5
Matched Problem 1 The matrices below are not in reduced form. Indi-
cate which condition in the definition is violated for each matrix. State the row
operation(s) required to transform the matrix into reduced form, and find the re-
duced form.
1 5 4 3
1 0 2 2 3
(A) c d (B) C 0 1 2 -1 S
0 3 -6
0 0 0 0
0 1 0 -3 1 2 0 3
(C) C 1 0 0 3 0S (D) C 0 0 0 3 0S
0 0 1 2 0 0 1 4
Solving Systems by Gauss–Jordan Elimination

We are now ready to outline the Gauss–Jordan method for solving systems of lin-
ear equations. The method systematically transforms an augmented matrix into a
reduced form. The system corresponding to a reduced augmented matrix is called a
reduced system. As we shall see, reduced systems are easy to solve.
The Gauss–Jordan elimination method is named after the German mathemati-
cian Carl Friedrich Gauss (1777–1885) and the German geodesist Wilhelm Jordan
(1842–1899). Gauss, one of the greatest mathematicians of all time, used a method of
solving systems of equations in his astronomical work that was later generalized by
Jordan to solve problems in large-scale surveying.
EXAMPLE 2 Solving a System Using Gauss–Jordan Elimination Solve by Gauss–Jordan

elimination:
2x1 - 2x2 + x3 = 3
3x1 + x2 - x3 = 7
x1 - 3x2 + 2x3 = 0
SOLUTION Write the augmented matrix and follow the steps indicated at the right.
R1 4 R3 Step 1 Choose the leftmost
"
2 -2 1 3
Need a 1 here. C3 1 -1 3 7S nonzero column and
get a 1 at the top.
1 -3 2 0
1 -3 2 0 Step 2 Use multiples of the row

C 3 1 -1 3 7S containing the 1 from
Need 0’s here. ∼ 1 - 32R1 + R2 S R2
step 1 to get zeros in all
"
2 -2 1 3 1 - 22R1 + R3 S R3
remaining places in the
" column containing this 1.
1 -3 2 0 Step 3 Repeat step 1 with

∼C 0 10 -7 3 7S 0.1R2 S R2 the submatrix formed by
(mentally) deleting the
"
Need a 1 here. 0 4 -3 3
top row.
"
1 -3 2 0 3R2 + R1 S R1 Step 4 Repeat step 2
Need 0’s here. ∼C 0 1 - 0.7 3 0.7 S with the entire ma-
0 4 -3 3 1 - 42R2 + R3 S R3 trix.
"
1 0 - 0.1 2.1 Step 3 Repeat step 1 with

3 the submatrix formed by
∼C 0 1 - 0.7 0.7 S 1 - 52R3 S R3
Need a 1 here. (mentally) deleting the
0 0 - 0.2 0.2
top rows.
"
0.1R3 + R1 S R1 Step 4 Repeat step 2

"
1 0 - 0.1 2.1
Need 0’s here. 0.7R3 + R2 S R2 with the entire ma-
3
"
C
∼ 0 1 - 0.7 0.7 S
trix.
0 0 1 -1
1 0 0 2 The matrix is now in reduced

£
∼ 0 1 0 3 0§ form, and we can solve the cor-
responding reduced system.
0 0 1 -1
x1 = 2
x2 = 0
x3 = - 1
The solution to this system is x1 = 2, x2 = 0, x3 = - 1. You should check this
solution in the original system.
Matched Problem 2 Solve by Gauss–Jordan elimination:

3x1 + x2 - 2x3 = 2
x1 - 2x2 + x3 = 3
2x1 - x2 - 3x3 = 3
PROCEDURE Gauss–Jordan Elimination

Step 1 Choose the leftmost nonzero column and use appropriate row operations
to get a 1 at the top.
Step 2 Use multiples of the row containing the 1 from step 1 to get zeros in all
remaining places in the column containing this 1.
Step 3 Repeat step 1 with the submatrix formed by (mentally) deleting the row
used in step 2 and all rows above this row.
Step 4 Repeat step 2 with the entire matrix, including the rows deleted mentally.
Continue this process until the entire matrix is in reduced form.
Note: If at any point in this process we obtain a row with all zeros to the left of
the vertical line and a nonzero number to the right, we can stop before we find
the reduced form since we will have a contradiction: 0 = n, n ∙ 0. We can then
conclude that the system has no solution.
Remarks
1. Even though each matrix has a unique reduced form, the sequence of steps pre-
sented here for transforming a matrix into a reduced form is not unique. For exam-
ple, it is possible to use row operations in such a way that computations involving
fractions are minimized. But we emphasize again that we are not interested in the
most efficient hand methods for transforming small matrices into reduced forms.
Our main interest is in giving you a little experience with a method that is suitable
for solving large-scale systems on a graphing calculator or computer.
2. Most graphing calculators have the ability to find reduced forms. Figure 1 illus-
trates the solution of Example 2 on a TI-84 Plus CE graphing calculator using the
rref command (rref is an acronym for reduced row echelon form). Notice that in
row 2 and column 4 of the reduced form the graphing calculator has displayed the
very small number –3.5E – 13, instead of the exact value 0. This is a common oc-
Figure 1 Gauss–Jordan elimination on currence on a graphing calculator and causes no problems. Just replace any very
a graphing calculator small numbers displayed in scientific notation with 0.

elimination:
2x1 - 4x2 + x3 = - 4
4x1 - 8x2 + 7x3 = 2
- 2x1 + 4x2 - 3x3 = 5
SOLUTION 2 -4 1 -4
C 4 -8 7 3 2 S 0.5R1 S R1
-2 4 -3 5
1 -2 0.5 -2
∼C 4 -8 7 3 2 S 1 - 42R1 + R2 S R2
-2 4 -3 5 2R1 + R3 S R3
1 -2 0.5 -2 0.2R2 S R2 Note that column 3 is the

3 10 S leftmost nonzero column
∼C 0 0 5
in this submatrix.
0 0 -2 1
1 -2 0.5 -2 1 - 0.52R2 + R1 S R1
∼ C 0 0 1 3 2 S
0 0 -2 1 2R2 + R3 S R3
1 -2 0 -3 We stop the Gauss–Jordan elimination,

∼C 0 0 1 3 2 S even though the matrix is not in
0 0 0 5 reduced form, since the last row
produces a contradiction.
The system has no solution.

2x1 - 4x2 - x3 = - 8
4x1 - 8x2 + 3x3 = 4
- 2x1 + 4x2 + x3 = 11
Figure 2 shows the solution to Example 3 on a graphing calcula-

! CAUTION tor with a built-in reduced-form routine. Notice that the graphing
calculator does not stop when a contradiction first occurs but continues on to find the
reduced form. Nevertheless, the last row in the reduced form still produces a contra-
diction. Do not confuse this type of reduced form with one that represents a consistent
system (see Fig. 1).
Figure 2 Recognizing contradictions on a graphing calculator

elimination:
3x1 + 6x2 - 9x3 = 15
2x1 + 4x2 - 6x3 = 10
- 2x1 - 3x2 + 4x3 = - 6
SOLUTION 3 6 -9 15
3 1
C 2 4 -6 10 S R S R1
3 1
-2 -3 4 -6
1 2 -3 5
3
∼C 2 4 -6 10 S 1 - 22R1 + R2 S R2
-2 -3 4 -6 2R1 + R3 S R3
Note that we must interchange
1 2 -3 5 rows 2 and 3 to obtain a
3
∼C 0 0 0 0S R2 4 R3 nonzero entry at the top of
0 1 -2 4 the second column of this
submatrix.
1 2 -3 5
3
∼C 0 1 -2 4S 1 - 22R2 + R1 S R1
0 0 0 0
1 0 1 -3 The matrix is now in reduced form.

3
∼C 0 1 -2 4S Write the corresponding reduced
0 0 0 0 system and solve.
All-zero rows x1 + x3 = - 3 We discard the equation corresponding

! CAUTION do not neces- to the third (all zero) row in the reduced
x2 - 2x3 = 4 form, since it is satisfied by all values of
sarily indicate that there are infinite-
ly many solutions. These rows show x1, x2, and x3.
that some of the information given Note that the leftmost variable in each equation appears in one and only one
by the equations was redundant. equation. We solve for the leftmost variables x1 and x2 in terms of the remaining
variable, x3:
x1 = - x3 - 3
x2 = 2x3 + 4
If we let x3 = t, then for any real number t,
x1 = - t - 3
x2 = 2t + 4
x3 = t
You should check that 1 - t - 3, 2t + 4, t2 is a solution of the original system for
any real number t. Some particular solutions are
t = 0 t = -2 t = 3.5
1 - 3, 4, 02 1 - 1, 0, - 22 1 - 6.5, 11, 3.52
More generally,
If the number of leftmost 1’s in a reduced augmented coefficient matrix is

less than the number of variables in the system and there are no contradic-
tions, then the system is dependent and has infinitely many solutions.
Describing the solution set to such a dependent system is not difficult. In a re-
duced system without contradictions, the leftmost variables correspond to the left-
most 1’s in the corresponding reduced augmented matrix. The definition of reduced
form for an augmented matrix ensures that each leftmost variable in the correspond-
ing reduced system appears in one and only one equation of the system. Solving for
each leftmost variable in terms of the remaining variables and writing a general solu-
tion to the system is usually easy. Example 5 illustrates a slightly more involved case.

2x1 - 2x2 - 4x3 = - 2
3x1 - 3x2 - 6x3 = - 3
- 2x1 + 3x2 + x3 = 7

Explain why the definition of reduced form ensures that each leftmost variable in a
reduced system appears in one and only one equation and no equation contains more
than one leftmost variable. Discuss methods for determining whether a consistent
system is independent or dependent by examining the reduced form.

elimination:
x1 + 2x2 + 4x3 + x4 - x5 = 1
2x1 + 4x2 + 8x3 + 3x4 - 4x5 = 2
x1 + 3x2 + 7x3 + 3x5 = - 2
SOLUTION 1 2 4 1 -1 1 1 - 22R1 + R2 S R2
C2 4 8 3 -4 3 2S 1 - 12R1 + R3 S R3
1 3 7 0 3 -2
1 2 4 1 -1 1
~C 0 0 0 1 -2 3 0S R2 4 R3
0 1 3 -1 4 -3
1 2 4 1 -1 1
~C 0 1 3 -1 4 3 -3 S 1 - 22R2 + R1 S R1
0 0 0 1 -2 0
1 0 -2 3 -9 7
~C 0 1 3 -1 4 3 -3 S 1 - 32R3 + R1 S R1
0 0 0 1 -2 0 R3 + R2 S R2
1 0 -2 0 -3 7
~C 0 1 3 0 2 3 -3 S Matrix is in reduced form.
0 0 0 1 -2 0
x1 - 2x3 - 3x5 = 7
x2 + 3x3 + 2x5 = - 3
x4 - 2x5 = 0
Solve for the leftmost variables x1, x2, and x4 in terms of the remaining variables
x3 and x5 :
x1 = 2x3 + 3x5 + 7
x2 = - 3x3 - 2x5 - 3
x4 = 2x5
If we let x3 = s and x5 = t, then for any real numbers s and t,
x1 = 2s + 3t + 7
x2 = - 3s - 2t - 3
x3 = s
x4 = 2t
x5 = t
is a solution. The check is left for you.

x1 - x2 + 2x3 - 2x5 = 3
- 2x1 + 2x2 - 4x3 - x4 + x5 = - 5
3x1 - 3x2 + 7x3 + x4 - 4x5 = 6
Application
Dependent systems of linear equations provide an excellent opportunity to discuss
mathematical modeling in more detail. The process of using mathematics to solve
real-world problems can be broken down into three steps (Fig. 3):
Real-world
problem
1.
ret
Co
erp
nst
Int
ruc
3.
t
Mathematical Mathematical
2. Solve
solution model
Figure 3
Step 1 Construct a mathematical model whose solution will provide information

about the real-world problem.
Step 2 Solve the mathematical model.
Step 3 Interpret the solution to the mathematical model in terms of the original real-
world problem.
In more complex problems, this cycle may have to be repeated several times to
obtain the required information about the real-world problem.
EXAMPLE 6 Purchasing A company that rents small moving trucks wants to purchase 25 trucks
with a combined capacity of 28,000 cubic feet. Three different types of trucks are
available: a 10-foot truck with a capacity of 350 cubic feet, a 14-foot truck with a
capacity of 700 cubic feet, and a 24-foot truck with a capacity of 1,400 cubic feet.
How many of each type of truck should the company purchase?
SOLUTION The question in this example indicates that the relevant variables are
the number of each type of truck:
x1 = number of 10-foot trucks
Next we form the mathematical model:
x1 + x2 + x3 = 25 Total number of trucks
(2)
350x1 + 700x2 + 1,400x3 = 28,000 Total capacity
Now we form the augmented matrix of the system and solve by Gauss–Jordan
elimination:
1 1 1 2
25 1
c d 350 R2 S R2
350 700 1,400 28,000
1 1 1 2 25
~c d - R1 + R2 S R2
1 2 4 80
1 1 1 2 25
~c d - R2 + R1 S R1
0 1 3 55
1 0 - 2 2 - 30
~c d Matrix is in reduced form.
0 1 3 55
x1 - 2x3 = - 30 or x1 = 2x3 - 30
x2 + 3x3 = 55 or x2 = - 3x3 + 55
Let x3 = t. Then for t any real number,

x1 = 2t - 30
x2 = - 3t + 55 (3)
x3 = t
is a solution to mathematical model (2).
Now we must interpret this solution in terms of the original problem. Since the
variables x1, x2, and x3 represent numbers of trucks, they must be nonnegative real
numbers. And since we can’t purchase a fractional number of trucks, each must be
a nonnegative whole number. Since t = x3, it follows that t must also be a nonnega-
tive whole number. The first and second equations in model (3) place additional
restrictions on the values that t can assume:
x1 = 2t - 30 Ú 0 implies that t Ú 15
55
x2 = - 3t + 55 Ú 0 implies that t … = 18 13
3
So the only possible values of t that will produce meaningful solutions to the original
problem are 15, 16, 17, and 18. That is, the only combinations of 25 trucks that will
result in a combined capacity of 28,000 cubic feet are x1 = 2t - 30 10-foot trucks,
x2 = - 3t + 55 14-foot trucks, and x3 = t 24-foot trucks, where t = 15, 16, 17, or
18. A table is a convenient way to display these solutions:
10-Foot Truck 14-Foot Truck 24-Foot Truck
t x1 x2 x3
15 0 10 15
16 2 7 16
17 4 4 17
18 6 1 18
Matched Problem 6 A company that rents small moving trucks wants to pur-
chase 16 trucks with a combined capacity of 19,200 cubic feet. Three different
types of trucks are available: a cargo van with a capacity of 300 cubic feet, a 15-
foot truck with a capacity of 900 cubic feet, and a 24-foot truck with a capacity of
1,500 cubic feet. How many of each type of truck should the company purchase?

Refer to Example 6. The rental company charges $19.95 per day for a 10-foot truck,
$29.95 per day for a 14-foot truck, and $39.95 per day for a 24-foot truck. Which of
the four possible choices in the table would produce the largest daily income from
truck rentals?
Exercises 3.3
Skills Warm-up Exercises 3. x1 + 6x3 = 2 4. 3x1 + 4x2 = 10
x2 - x3 = 5 x1 + 5x3 = 15
W In Problems 1–4, write the augmented matrix of the system of
linear equations. (If necessary, review the terminology of Section x1 + 3x2 = 7 - x2 + x3 = 20
3.2.) In Problems 5–8, write the system of linear equations that is
1. x1 + 2x2 + 3x3 = 12 2. 4x1 + x2 = 8 represented by the augmented matrix. Assume that the variables
x1 + 7x2 - 5x3 = 15 3x1 - 5x2 = 6 are x1, x2, c .
x1 + 9x2 = 4 1 -3 4
-1 8 -3 0
5. C 3 2 3 5S 6. c 2 d
9 -2 0 8
-1 6 3
1 0 -1 1 32. In which of Problems 20, 22, 24, 26, and 28 is the number of
8. £ - 1 3 3§ leftmost ones less than the number of variables?
7. 35 -2 0 8 ∙ 44 1 0
0 2 1 -5 In Problems 33–38, discuss the validity of each statement about
linear systems. If the statement is always true, explain why. If not,
A In Problems 9–18, if a matrix is in reduced form, say so. If not, give a counterexample.
explain why and indicate a row operation that completes the next 33. If the number of leftmost ones is equal to the number of vari-
step of Gauss–Jordan elimination. ables, then the system has exactly one solution.
1 0 3 0 1 3 34. If the number of leftmost ones is less than the number of vari-
9. c 2 d 10. c 2 d
0 1 -2 1 0 -2 ables, then the system has infinitely many solutions.
35. If the number of leftmost ones is equal to the number of
0 1 5 1 0 5
11. c 2 d 12. c 2 d variables and the system is consistent, then the system has
1 0 -1 0 1 -1
exactly one solution.
4 12 -8 16 5 10 -5 - 15 36. If the number of leftmost ones is less than the number of
13. C 0 3 -6 3 0S 14. C 0 2 -2 3 7S variables and the system is consistent, then the system has
0 7 -2 6 0 5 -1 0 infinitely many solutions.
1 5 -7 2 1 10 -5 - 15 37. If there is an all-zero row, then the system has infinitely many
3 3 solutions.
15. C 0 1 -6 0S 16. C 0 0 -2 6S
0 7 -2 6 0 0 0 0 38. If there are no all-zero rows, then the system has exactly one
solution.
1 2 2 9 1 0 -5 - 15
17. C 0 0 -2 3 8S 18. C 0 1 -2 3 7S B Use row operations to change each matrix in Problems 39–46 to
0 0 0 0 0 5 -1 0 reduced form.
In Problems 19–28, write the solution of the linear system cor- 1 2 -1 1 3 1

39. c 2 d 40. c 2 d
responding to each reduced augmented matrix. 0 1 3 0 2 -4
1 0 0 0 -2 1 1 1 2 16 1 1 1 2 8
1 0 0 -2 41. c d 42. c d
0 1 0 0 0 2 3 4 25 3 5 7 30
19. C 0 1 0 3 3S 20. D 4 T
0 0 1 0 1 1 0 -3 1 1 0 4 0
0 0 1 0
0 0 0 1 3 43. C 0 1 2 3 0S 44. C 0 1 -3 3 -1 S
1 0 -2 3 1 -2 0 -3 0 0 3 -6 0 0 -2 2
21. C 0 1 1 3 -5 S 22. C 0 0 1 3 5S 1 2 -2 -1 1 -2 3 -5
0 0 0 0 0 0 0 0 45. C 0 3 -6 3 -3 S 46. C 0 4 -8 3 -4 S
0 -1 2 1 0 -1 2 1
1 0 0 1 0 7
23. C 0 1 3 0S 24. C 0 1 3 -8 S Solve Problems 47–62 using Gauss–Jordan elimination.
0 0 1 0 0 0
47. 2x1 + 4x2 - 10x3 = - 2 48. 3x1 + 5x2 - x3 = - 7
1 0 -3 5 1 0 1 -4 3x1 + 9x2 - 21x3 = 0 x1 + x2 + x3 = - 1
25. c 2 d 26. c 2 d
0 1 2 -7 0 1 -1 6 x1 + 5x2 - 12x3 = 1 2x1 + 11x3 = 7
49. 3x1 + 8x2 - x3 = - 18 50. 2x1 + 6x2 + 15x3 = - 12
1 -2 0 -3 -5
27. c 2 d 2x1 + x2 + 5x3 = 8 4x1 + 7x2 + 13x3 = - 10
0 0 1 3 2
2x1 + 4x2 + 2x3 = - 4 3x1 + 6x2 + 12x3 = - 9
1 0 -2 3 4 51. 2x1 - x2 - 3x3 = 8 52. 2x1 + 4x2 - 6x3 = 10
28. c 2 d
0 1 -1 2 -1 x1 - 2x2 = 7 3x1 + 3x2 - 3x3 = 6
53. 2x1 - x2 = 0 54. 2x1 - x2 = 0
29. In which of Problems 19, 21, 23, 25, and 27 is the number of
3x1 + 2x2 = 7 3x1 + 2x2 = 7
leftmost ones equal to the number of variables?
x1 - x2 = -1 x1 - x2 = -2
30. In which of Problems 20, 22, 24, 26, and 28 is the number of 55. 3x1 - 4x2 - x3 = 1 56. 3x1 + 7x2 - x3 = 11
leftmost ones equal to the number of variables?
2x1 - 3x2 + x3 = 1 x1 + 2x2 - x3 = 3
31. In which of Problems 19, 21, 23, 25, and 27 is the number of x1 - 2x2 + 3x3 = 2 2x1 + 4x2 - 2x3 = 10
leftmost ones less than the number of variables?
57. 3x1 - 2x2 + x3 = - 7 58. 2x1 + 3x2 + 5x3 = 21

2x1 + x2 - 4x3 = 0 x1 - x2 - 5x3 = - 2 Applications
x1 + x2 - 3x3 = 1 2x1 + x2 - x3 = 11 Construct a mathematical model for each of the following
59. 2x1 + 4x2 - 2x3 = 2 60. 4x1 - 16x2 + 8x3 = - 20 problems. (The answers in the back of the book include both the
- 3x1 - 6x2 + 3x3 = - 3 - 5x1 + 20x2 - 10x3 = 25 mathematical model and the interpretation of its solution.) Use
61. 4x1 - x2 + 2x3 = 3 62. 4x1 - 2x2 + 2x3 = 5 Gauss–Jordan elimination to solve the model and then interpret
the solution.
- 4x1 + x2 - 3x3 = - 10 - 6x1 + 3x2 - 3x3 = - 2
8x1 - 2x2 + 9x3 = - 1 10x1 - 5x2 + 9x3 = 4 73. Boat production. A small manufacturing plant makes
63. Consider a consistent system of three linear equations in three types of inflatable boats: one-person, two-person, and
three variables. Discuss the nature of the system and its four-person models. Each boat requires the services of three
solution set if the reduced form of the augmented coefficient departments, as listed in the table. The cutting, assembly, and
matrix has packaging departments have available a maximum of 380,
330, and 120 labor-hours per week, respectively.
(A) One leftmost 1 (B) Two leftmost 1’s
(C) Three leftmost 1’s (D) Four leftmost 1’s One-Person Two-Person Four-Person
Department Boat Boat Boat
64. Consider a system of three linear equations in three variables.
Cutting 0.5 hr 1.0 hr 1.5 hr
Give examples of two reduced forms that are not row-equiva-
lent if the system is Assembly 0.6 hr 0.9 hr 1.2 hr
(A) Consistent and dependent Packaging 0.2 hr 0.3 hr 0.5 hr
(B) Inconsistent (A) How many boats of each type must be produced each
week for the plant to operate at full capacity?
C Solve Problems 65–70 using Gauss–Jordan elimination.
(B) How is the production schedule in part (A) affected if
65. x1 + 2x2 - 4x3 - x4 = 7
the packaging department is no longer used?
2x1 + 5x2 - 9x3 - 4x4 = 16
x1 + 5x2 - 7x3 - 7x4 = 13 (C) How is the production schedule in part (A) affected if
66. 2x1 + 4x2 + 5x3 + 4x4 = 8 the four-person boat is no longer produced?
x1 + 2x2 + 2x3 + x4 = 3 74. Production scheduling. Repeat Problem 73 assuming that
67. x1 - x2 + 3x3 - 2x4 = 1 the cutting, assembly, and packaging departments have avail-
able a maximum of 350, 330, and 115 labor-hours per week,
- 2x1 + 4x2 - 3x3 + x4 = 0.5
respectively.
3x1 - x2 + 10x3 - 4x4 = 2.9
4x1 - 3x2 + 8x3 - 2x4 = 0.6 75. Tank car leases. A chemical manufacturer wants to lease a
68. x1 + x2 + 4x3 + x4 = 1.3 fleet of 24 railroad tank cars with a combined carrying capacity
of 520,000 gallons. Tank cars with three different carrying ca-
- x1 + x2 - x3 = 1.1
pacities are available: 8,000 gallons, 16,000 gallons, and 24,000
2x1 + x3 + 3x4 = - 4.4
gallons. How many of each type of tank car should be leased?
2x1 + 5x2 + 11x3 + 3x4 = 5.6
69. x1 - 2x2 + x3 + x4 + 2x5 = 2 76. Airplane leases. A corporation wants to lease a fleet of
12 airplanes with a combined carrying capacity of 220 pas-
- 2x1 + 4x2 + 2x3 + 2x4 - 2x5 = 0
sengers. The three available types of planes carry 10, 15, and
3x1 - 6x2 + x3 + x4 + 5x5 = 4
20 passengers, respectively. How many of each type of plane
- x1 + 2x2 + 3x3 + x4 + x5 = 3
should be leased?
70. x1 - 3x2 + x3 + x4 + 2x5 = 2
77. Tank car leases. Refer to Problem 75. The cost of leasing
- x1 + 5x2 + 2x3 + 2x4 - 2x5 = 0
an 8,000-gallon tank car is $450 per month, a 16,000-gal-
2x1 - 6x2 + 2x3 + 2x4 + 4x5 = 4
lon tank car is $650 per month, and a 24,000-gallon tank car
- x1 + 3x2 - x3 + x5 = - 3
is $1,150 per month. Which of the solutions to Problem 75
71. Find a, b, and c so that the graph of the quadratic equa- would minimize the monthly leasing cost?
tion y = ax2 + bx + c passes through the points 1 - 2, 92,
11, - 92, and 14, 92. 78. Airplane leases. Refer to Problem 76. The cost of leasing
a 10-passenger airplane is $8,000 per month, a 15-passenger
72. Find a, b, and c so that the graph of the quadratic equation airplane is $14,000 per month, and a 20-passenger airplane
y = ax2 + bx + c passes through the points 1 - 1, - 52, is $16,000 per month. Which of the solutions to Problem 76
12, 72, and 15, 12. would minimize the monthly leasing cost?
79. Income tax. A corporation has a taxable income of 84. Nutrition. Repeat Problem 83 if the diet is to include exactly
$7,650,000. At this income level, the federal income tax rate 400 units of calcium, 160 units of iron, and 240 units of
is 50%, the state tax rate is 20%, and the local tax rate is vitamin A.
10%. If each tax rate is applied to the total taxable income,
85. Plant food. A farmer can buy four types of plant food. Each
the resulting tax liability for the corporation would be 80% of
barrel of mix A contains 30 pounds of phosphoric acid, 50
taxable income. However, it is customary to deduct taxes paid
pounds of nitrogen, and 30 pounds of potash; each barrel of
to one agency before computing taxes for the other agencies.
mix B contains 30 pounds of phosphoric acid, 75 pounds of
Assume that the federal taxes are based on the income that
nitrogen, and 20 pounds of potash; each barrel of mix C con-
remains after the state and local taxes are deducted, and that
tains 30 pounds of phosphoric acid, 25 pounds of nitrogen,
state and local taxes are computed in a similar manner. What
and 20 pounds of potash; and each barrel of mix D contains
is the tax liability of the corporation (as a percentage of tax-
60 pounds of phosphoric acid, 25 pounds of nitrogen, and
able income) if these deductions are taken into consideration?
50 pounds of potash. Soil tests indicate that a particular field
80. Income tax. Repeat Problem 79 if local taxes are not al- needs 900 pounds of phosphoric acid, 750 pounds of nitro-
lowed as a deduction for federal and state taxes. gen, and 700 pounds of potash. How many barrels of each
type of food should the farmer mix together to supply the
81. Taxable income. As a result of several mergers and acquisi-
necessary nutrients for the field?
tions, stock in four companies has been distributed among
the companies. Each row of the following table gives the 86. Animal feed. In a laboratory experiment, rats are to be fed
percentage of stock in the four companies that a particular 5 packets of food containing a total of 80 units of vitamin E.
company owns and the annual net income of each company There are four different brands of food packets that can be
(in millions of dollars): used. A packet of brand A contains 5 units of vitamin E, a
Percentage of Stock Annual Net packet of brand B contains 10 units of vitamin E, a packet of
Owned in Company Income brand C contains 15 units of vitamin E, and a packet of brand
Company A B C D Million $ D contains 20 units of vitamin E. How many packets of each
brand should be mixed and fed to the rats?
A 71 8 3 7 3.2
B 12 81 11 13 2.6 87. Plant food. Refer to Problem 85. The costs of the four mixes
C 11 9 72 8 3.8
are Mix A, $46; Mix B, $72; Mix C, $57; and Mix D, $63.
Which of the solutions to Problem 85 would minimize the
D 6 2 14 72 4.4
cost of the plant food?
So company A holds 71% of its own stock, 8% of the stock 88. Animal feed. Refer to Problem 86. The costs of the four
in company B, 3% of the stock in company C, etc. For the brands are Brand A, $1.50; Brand B, $3.00; Brand C, $3.75;
purpose of assessing a state tax on corporate income, the tax- and Brand D, $2.25. Which of the solutions to Problem 86
able income of each company is defined to be its share of its would minimize the cost of the rat food?
own annual net income plus its share of the taxable income of
89. Population growth. The population of Spain was
each of the other companies, as determined by the percentag-
approximately 34 million in 1970, 39 million in 1990,
es in the table. What is the taxable income of each company
and 46 million in 2010. Construct a model for this data by
(to the nearest thousand dollars)?
finding a quadratic equation whose graph passes through
82. Taxable income. Repeat Problem 81 if tax law is changed the points (0,34), (20,39), and (40,46). Use this model to
so that the taxable income of a company is defined to be all estimate the population in 2030.
of its own annual net income plus its share of the taxable
income of each of the other companies. 90. Population growth. The population of Turkey was ap-
proximately 35 million in 1970, 56 million in 1990, and
83. Nutrition. A dietitian in a hospital is to arrange a special 73 million in 2010. Construct a model for this data by finding
diet composed of three basic foods. The diet is to include a quadratic equation whose graph passes through the points
exactly 340 units of calcium, 180 units of iron, and 220 units (0,35), (20,56), and (40,73). Use this model to estimate the
of vitamin A. The number of units per ounce of each special population in 2030.
ingredient for each of the foods is indicated in the table.
91. Female life expectancy. The life expectancy for fe-
Units per Ounce males born during 1980–1985 was approximately 77.6
Food A Food B Food C years. This grew to 78 years during 1985–1990 and to
Calcium 30 10 20 78.6 years during 1990–1995. Construct a model for this
Iron 10 10 20 data by finding a quadratic equation whose graph passes
through the points (0, 77.6), (5, 78), and (10, 78.6). Use
Vitamin A 10 30 20
this model to estimate the life expectancy for females
(A) How many ounces of each food must be used to meet the born between 1995 and 2000 and for those born between
diet requirements? 2000 and 2005.
(B) How is the diet in part (A) affected if food C is not used? 92. Male life expectancy. The life expectancy for males born
during 1980–1985 was approximately 70.7 years. This grew
(C) How is the diet in part (A) affected if the vitamin A re- to 71.1 years during 1985–1990 and to 71.8 years during
quirement is dropped? 1990–1995. Construct a model for this data by finding a
quadratic equation whose graph passes through the points (D) If traffic lights are adjusted so that 600 vehicles per
(0, 70.7), (5, 71.1), and (10, 71.8). Use this model to estimate hour travel from Baker Street to Market Street on
the life expectancy for males born between 1995 and 2000 Old Street, determine the flow around the rest of the
and for those born between 2000 and 2005. network.
93. Female life expectancy. Refer to Problem 91. Subsequent 98. Traffic flow. Refer to Problem 97. Closing Bridge Street
data indicated that life expectancy grew to 79.1 years for north of Baker Street for construction changes the traffic
females born during 1995–2000 and to 79.7 years for females flow for the network as indicated in the figure. Repeat
born during 2000–2005. Add the points (15, 79.1) and (20, 79.7) parts (A)–(D) of Problem 97 for this traffic flow.
to the data set in Problem 91. Use a graphing calculator to find
a quadratic regression model for all five data points. Graph the Old St. Bridge St.
data and the model in the same viewing window.
500
94. Male life expectancy. Refer to Problem 92. Subsequent
data indicated that life expectancy grew to 73.2 years for 700 x1 700
Baker St.
males born during 1995–2000 and to 74.3 years for males x4 x2
born during 2000–2005. Add the points (15, 73.2) and (20,
74.3) to the data set in Problem 92. Use a graphing calculator Market St.
800 x3 350
to find a quadratic regression model for all five data points. 600 550
Graph the data and the model in the same viewing window.
95. Sociology. Two sociologists have grant money to study
school busing in a particular city. They wish to conduct an Answers to Matched Problems
opinion survey using 600 telephone contacts and 400 house
contacts. Survey company A has personnel to do 30 tele- 1. (A) Condition 2 is violated: The 3 in row 2 and column 2
phone and 10 house contacts per hour; survey company B can should be a 1. Perform the operation 13 R2 S R2 to obtain
handle 20 telephone and 20 house contacts per hour. How 1 0 2
c 2 d
many hours should be scheduled for each firm to produce 0 1 -2
exactly the number of contacts needed?
(B) Condition 3 is violated: The 5 in row 1 and col-
96. Sociology. Repeat Problem 95 if 650 telephone contacts and umn 2 should be a 0. Perform the operation
350 house contacts are needed. 1 - 52R2 + R1 S R1 to obtain
97. Traffic flow. The rush-hour traffic flow for a network of four
1 0 -6 8
one-way streets in a city is shown in the figure. The numbers
C0 1 2 3 -1 S
next to each street indicate the number of vehicles per hour
that enter and leave the network on that street. The variables 0 0 0 0
x1, x2, x3, and x4 represent the flow of traffic between the four
intersections in the network. (C) Condition 4 is violated. The leftmost 1 in the second
row is not to the right of the leftmost 1 in the first row.
(A) For a smooth traffic flow, the number of vehicles enter- Perform the operation R1 4 R2 to obtain
ing each intersection should always equal the number
leaving. For example, since 1,200 vehicles enter the 1 0 0 0
intersection of Old Street and Baker Street each hour C0 1 0 3 -3 S
and x1 + x4 vehicles leave this intersection, we see that 0 0 1 2
x1 + x4 = 1,200. Find the equations determined by the
traffic flow at each of the other three intersections. (D) Condition 1 is violated: The all-zero second row should
be at the bottom. Perform the operation R2 4 R3 to
Old St. Bridge St. obtain
500 450 1 2 0 3
C0 0 1 3 4S
Baker St.
700 x1 550
0 0 0 0
x4 x2
Market St. 2. x1 = 1, x2 = - 1, x3 = 0
600 x3 350
3. Inconsistent; no solution
500 550
4. x1 = 5t + 4, x2 = 3t + 5, x3 = t, t any real number
5. x1 = s + 7, x2 = s, x3 = t - 2, x4 = - 3t - 1, x5 = t, s
(B) Find the solution to the system in part (A).
and t any real numbers
(C) What is the maximum number of vehicles that can travel 6. t - 8 cargo vans, - 2t + 24 15-foot trucks, and t 24-foot
from Baker Street to Market Street on Old Street? What trucks, where t = 8, 9, 10, 11, or 12
is the minimum number?
SECTION 3.4 Matrices: Basic Operations 189
3.4 Matrices: Basic Operations

■■ Addition and Subtraction In the two preceding sections we introduced the important idea of matrices. In this
■■ Product of a Number k and
and following sections, we develop this concept further. Matrices are both an ancient
a Matrix M and a current mathematical concept. References to matrices and systems of equations
can be found in Chinese manuscripts dating back to about 200 b.c. More recently,
■■ Matrix Product computers have made matrices a useful tool for a wide variety of applications. Most
graphing calculators and computers are capable of performing calculations with
matrices.
As we will see, matrix addition and multiplication are similar to real number ad-
dition and multiplication in many respects, but there are some important differences.
A brief review of Appendix B, Section B.1, where real number operations are dis-
cussed, will help you understand the similarities and the differences.
Addition and Subtraction

Before we can discuss arithmetic operations for matrices, we have to define equality
for matrices. Two matrices are equal if they have the same size and their correspond-
ing elements are equal. For example,
2 * 3 2 * 3
a b c u v w a = u b = v c = w
c d = c d if and only if
d e f x y z d = x e = y f = z
The sum of two matrices of the same size is the matrix with elements that are
the sum of the corresponding elements of the two given matrices. Addition is not de-
fined for matrices of different sizes.
EXAMPLE 1 Matrix Addition

a b w x 1a + w2 1b + x2
(A) c d + c d = c d
c d y z 1c + y 2 1d + z2
2 -3 0 3 1 2 5 -2 2
(B) c d + c d = c d
1 2 -5 -3 2 5 -2 4 0
-1 7
5 0 -2
(C) c d + C 0 6S Not defined
1 -3 8
-2 8
3 2 -2 3
Matched Problem 1 Add: C - 1 -1 S + C 1 -1 S
0 3 2 -2
Graphing calculators can be used to solve problems involving matrix operations.

Figure 1 illustrates the solution to Example 1B on a TI-84 Plus CE.
Because we add two matrices by adding their corresponding elements, it follows
from the properties of real numbers that matrices of the same size are commutative
and associative relative to addition. That is, if A, B, and C are matrices of the same
size, then
Commutative: A + B = B + A
Figure 1 Addition on a graphing
calculator Associative: 1A + B2 + C = A + 1B + C2
A matrix with elements that are all zeros is called a zero matrix. For example,
[0 0 0 0] 0
0 0 0 0 0
c d
0 0
C0 0 0 0S 0
D T
0 0 0 0 0
0
are zero matrices of different sizes. [Note: The simpler notation “0” is often used to
denote the zero matrix of an arbitrary size.] The negative of a matrix M, denoted by
∙ M, is a matrix with elements that are the negatives of the elements in M. Thus, if
a b -a -b
M = c d then -M = c d
c d -c -d
Note that M + 1 - M2 = 0 (a zero matrix).
If A and B are matrices of the same size, we define subtraction as follows:

A ∙ B = A ∙ 1 ∙ B2
So to subtract matrix B from matrix A, we simply add the negative of B to A.
EXAMPLE 2 Matrix Subtraction
3 -2 -2 2 3 -2 2 -2 5 -4
c d - c d = c d + c d = c d
5 0 3 4 5 0 -3 -4 2 -4
Matched Problem 2 Subtract: [2 -3 5 ] - [3 -2 1]
EXAMPLE 3 Matrix Equations Find a, b, c, and d so that
a b 2 -1 4 3
c d - c d = c d
c d -5 6 -2 4
SOLUTION a b 2 -1 4 3 Subtract the matrices on

c d - c d = c d
c d -5 6 -2 4 the left side.
a - 2 b - 1 - 12 4 3
c d = c d Remove parentheses.
c - 1 - 52 d - 6 -2 4
a - 2 b + 1 4 3 Use the definition of equality
c d = c d to change this matrix equation
c + 5 d - 6 -2 4
into four real number equations.
a - 2 = 4 b + 1 = 3 c + 5 = -2 d - 6 = 4
a = 6 b = 2 c = -7 d = 10
Matched Problem 3 Find a, b, c, and d so that
a b -4 2 -2 5
c d - c d = c d
c d 1 -3 8 2
Product of a Number k and a Matrix M

The product of a number k and a matrix M, denoted by kM, is a matrix formed by
multiplying each element of M by k.
EXAMPLE 4 Multiplication of a Matrix by a Number
3 -1 0 -6 2 0
-2 C -2 1 3S = C 4 -2 -6 S
0 -1 -2 0 2 4
1.3
Matched Problem 4 Find: 10 C 0.2 S
3.5
The next example illustrates the use of matrix operations in an applied setting.
EXAMPLE 5 Sales Commissions Ms. Smith and Mr. Jones are salespeople in a new-car agency
that sells only two models. August was the last month for this year’s models, and
next year’s models were introduced in September. Gross dollar sales for each month
are given in the following matrices:
August sales September sales
Compact Luxury Compact Luxury
Ms. Smith $54,000 $88,000 $228,000 $368,000
Mr. Jones
c d = A c d = B
$126,000 0 $304,000 $322,000
For example, Ms. Smith had $54,000 in compact sales in August, and
Mr. Jones had $322,000 in luxury car sales in September.
(A) What were the combined dollar sales in August and September for each sales-
person and each model?
(B) What was the increase in dollar sales from August to September?
(C) If both salespeople receive 5% commissions on gross dollar sales, compute the
commission for each person for each model sold in September.
SOLUTION
Compact Luxury
$282,000 $456,000 Ms. Smith
(A) A + B = c d
$430,000 $322,000 Mr. Jones
Compact Luxury
$174,000 $280,000 Ms. Smith
(B) B - A = c d
$178,000 $322,000 Mr. Jones
10.0521$228,0002 10.0521$368,0002
(C) 0.05B = c d
10.0521$304,0002 10.0521$322,0002
$11,400 $18,400 Ms. Smith

= c d Mr. Jones
$15,200 $16,100
Matched Problem 5 Repeat Example 5 with
$45,000 $77,000 $190,000 $345,000

A = c d and B = c d
$106,000 $22,000 $266,000 $276,000
Figure 2 illustrates a solution for Example 5 on a spreadsheet.
Figure 2
Matrix Product
Matrix multiplication was introduced by the English mathematician Arthur Cayley
(1821–1895) in studies of systems of linear equations and linear transformations.
Although this multiplication may seem strange at first, it is extremely useful in many
practical problems.
We start by defining the product of two special matrices, a row matrix and a col-
umn matrix.
DEFINITION Product of a Row Matrix and a Column Matrix

The product of a 1 * n row matrix and an n * 1 column matrix is a 1 * 1
matrix given by
n * 1
b1
1 * nb
3a1 a2 c an 4 D 2 T = 3a1b1 + a2b2 + g + anbn 4
f
bn
Note that the number of elements in the row matrix and in the column matrix
must be the same for the product to be defined.
EXAMPLE 6 Product of a Row Matrix and a Column Matrix
-5
[2 -3 0 ] C 2 S = 31221 - 52 + 1 - 32122 + 1021 - 224
-2
= 3 - 10 - 6 + 04 = 3 - 164
2
3
Matched Problem 6 [- 1 0 3 2]D T = ?
4
-1
Refer to Example 6. The distinction between the real number - 16 and the
1 * 1 matrix [- 16] is a technical one, and it is common to see 1 * 1 matrices writ-
ten as real numbers without brackets. In the work that follows, we will frequently
refer to 1 * 1 matrices as real numbers and omit the brackets whenever it is conve-
nient to do so.
EXAMPLE 7 Labor Costs A factory produces a slalom water ski that requires 3 labor-hours in
the assembly department and 1 labor-hour in the finishing department. Assembly
personnel receive $9 per hour and finishing personnel receive $6 per hour. Total
labor cost per ski is given by the product:
9
33 14 c d = 3132192 + 1121624 = 327 + 64 = 3334 or $33 per ski
6
Matched Problem 7 If the factory in Example 7 also produces a trick water

ski that requires 5 labor-hours in the assembly department and 1.5 labor-hours in
the finishing department, write a product between appropriate row and column
matrices that will give the total labor cost for this ski. Compute the cost.
We now use the product of a 1 * n row matrix and an n * 1 column matrix to

extend the definition of matrix product to more general matrices.
DEFINITION Matrix Product

If A is an m * p matrix and B is a p * n matrix, then the matrix product of A
and B, denoted AB, is an m * n matrix whose element in the ith row and jth col-
umn is the real number obtained from the product of the ith row of A and the jth
column of B. If the number of columns in A does not equal the number of rows in
B, the matrix product AB is not defined.
Must be the same It is important to check sizes before starting the multiplication process. If A is an
(b 5 c) Size of a * b matrix and B is a c * d matrix, then if b = c, the product AB will exist and will
product be an a * d matrix (see Fig. 3). If b ∙ c, the product AB does not exist. The definition
a3b c3d a3d
is not as complicated as it might first seem. An example should help clarify the process.
A ? B 5 AB
For
Figure 3 1 3
2 3 -1
A = c d and B = C 2 0S
-2 1 2
-1 2
A is 2 * 3 and B is 3 * 2, so AB is 2 * 2. To find the first row of AB, we take the

product of the first row of A with every column of B and write each result as a real
number, not as a 1 * 1 matrix. The second row of AB is computed in the same man-
ner. The four products of row and column matrices used to produce the four elements
in AB are shown in the following dashed box. These products are usually calculated
mentally or with the aid of a calculator, and need not be written out. The shaded
portions highlight the steps involved in computing the element in the first row and
second column of AB.
1 3
3 * 2 32 3 - 14 C 2 S 32 3 - 14 C 0 S
2 * 3
1 3 -1 2
2 3 -1
c dC 2 0S = G W
-2 1 2 1 3
-1 2 3 -2 1 24 C 2 S 3 -2 1 24 C 0 S
-1 2
2 * 2
122112 + 132122 + 1 - 121 - 12 122132 + 132102 + 1 - 12122 9 4
= c d = c d
1 - 22112 + 112122 + 1221 - 12 1 - 22132 + 112102 + 122122 -2 -2
EXAMPLE 8 Matrix Multiplication Find the indicated matrix product, if it exists, where:
2 1
1 -1 0 1 2 6
A = £ 1 0§ B = c d C = c d
2 1 2 0 -1 -3
-1 2
-5
1 2
D = c d E = 32 -3 04 F = £ 2§
3 6
-2
3 * 2 2 * 4
2 1
1 -1 0 1
(A) AB = £ 1 0§ c d
2 1 2 0
-1 2
122112 + 112122 1221 - 12 + 112112 122102 + 112122 122112 + 112102

= £ 112112 + 102122 1121 - 12 + 102112 112102 + 102122 112112 + 102102 §
1 - 12112 + 122122 1 - 121 - 12 + 122112 1 - 12102 + 122122 1 - 12112 + 122102
3 * 4
4 -1 2 2
= £1 -1 0 1§
3 3 4 -1
2 * 4 3 * 2
2 1
1 -1 0 1
(B) BA = c d£ 1 0§ Not defined
2 1 2 0
-1 2
2 6 1 2 122112 + 162132 122122 + 162162

(C) CD = c dc d = c d
-1 -3 3 6 1 -12112 + 1 - 32132 1 - 12122 + 1 - 32162
20 40
= c d
- 10 - 20
1 2 2 6 112122 + 1221 - 12 112162 + 1221 - 32

(D) DC = c dc d = c d
3 6 -1 -3 132122 + 1621 - 12 132162 + 1621 - 32
0 0
= c d
0 0
-5
(E) EF = 32 -3 04 C 2 S = 31221 - 52 + 1 - 32122 + 1021 - 224 = [- 16]
-2
-5 1 - 52122 1 - 521 - 32 1 - 52102

(F) FE = C 2 S [2 -3 0] = C 122122 1221 - 32 122102 S
-2 1 - 22122 1 - 221 - 32 1 - 22102
- 10 15 0
= C 4 -6 0S
-4 6 0
3 * 2 3 * 2
2 1 2 1
2
(G) A * = AA = £ 1 0§ £ 1 0§ Not defined
-1 2 -1 2
2 6 2 6
(H) C 2 = CC = c dc d
-1 -3 -1 -3
122122 + 1621 - 12 122162 + 1621 - 32
= c d
1 -12122 + 1 - 321 - 12 1 - 12162 + 1 - 321 - 32
-2 -6
= c d
1 3
Matched Problem 8 Find each product, if it is defined:
-1 1 -1 1
-1 0 3 -2 -1 0 3 -2
(A) c dC 2 3S (B) C 2 3S c d
1 2 2 0 1 2 2 0
1 0 1 0
1 2 -2 4 -2 4 1 2
(C) c dc d (D) c dc d
-1 -2 1 -2 1 -2 -1 -2
4 4
(E) [3 -2 1] C 2 S (F) C 2 S [3 -2 1]
3 3
Figure 4 illustrates a graphing calculator solution to Example 8A. What would

you expect to happen if you tried to solve Example 8B on a graphing calculator?
CONCEPTUAL INSIGHT
Figure 4 Multiplication on a In the arithmetic of real numbers, it does not matter in which order we multiply.
graphing calculator
For example, 5 * 7 = 7 * 5. In matrix multiplication, however, it does make a
difference. That is, AB does not always equal BA, even if both multiplications are
defined and both products are the same size (see Examples 8C and 8D).
Matrix multiplication is not commutative.
The zero property of real numbers states that if the product of two real numbers is
0, then one of the numbers must be 0 (see Appendix B, Section B.1). This property
is very important when solving equations. For example,
(Continued)
*Following standard algebraic notation, we write A2 = AA, A3 = AAA, and so on.

x2 - 4x + 3 = 0
1x - 121x - 32 = 0
x - 1 = 0 or x - 3 = 0
x = 1 x = 3
For matrices, it is possible to find nonzero matrices A and B such that AB is a zero
matrix (see Example 8D).
The zero property does not hold for matrix multiplication.

In addition to the commutative and zero properties, there are other significant differ-
ences between real number multiplication and matrix multiplication.
(A) In real number multiplication, the only real number whose square is 0 is the real
number 0 102 = 02. Find at least one 2 * 2 matrix A with all elements nonzero
such that A2 = 0, where 0 is the 2 * 2 zero matrix.
(B) In real number multiplication, the only nonzero real number that is equal to its
square is the real number 1 112 = 12. Find at least one 2 * 2 matrix B with all
elements nonzero such that B2 = B.
EXAMPLE 9 Matrix Multiplication Find a, b, c, and d so that
2 -1 a b -6 17
c dc d = c d
5 3 c d 7 4
SOLUTION The product of the matrices on the left side of the equation is
2 -1 a b 2a - c 2b - d
c dc d = c d
5 3 c d 5a + 3c 5b + 3d
Therefore,
2a - c = - 6 2b - d = 17
5a + 3c = 7 5b + 3d = 4
This gives a system of two equations in the variables a and c, and a second system
of two equations in the variables b and d. Each system can be solved by substitution,
or elimination by addition, or Gauss–Jordan elimination (the details are omitted).
The solution of the first system is a = - 1, c = 4, and the solution of the second
system is b = 5, d = - 7.
Matched Problem 9 Find a, b, c, and d so that
6 -5 a b - 16 64
c dc d = c d
0 3 c d 24 -6
Now we consider an application of matrix multiplication.
EXAMPLE 10 Labor Costs We can combine the time requirements for slalom and trick water
skis discussed in Example 7 and Matched Problem 7 into one matrix:
Labor-hours per ski
Assembly Finishing
department department
Trick ski 5 hr 1.5 hr
c d = L
Slalom ski 3 hr 1 hr
Now suppose that the company has two manufacturing plants, one in California
and the other in Maryland, and that their hourly rates for each department are given
in the following matrix:
Hourly wages
California Maryland
Assembly department $12 $13
c d = H
Finishing department $7 $8
Since H and L are both 2 * 2 matrices, we can take the product of H and L in
either order and the result will be a 2 * 2 matrix:
12 13 5 1.5 99 31
HL = c dc d = c d
7 8 3 1 59 18.5
5 1.5 12 13 70.5 77
LH = c dc d = c d
3 1 7 8 43 47
How can we interpret the elements in these products? Let’s begin with the product
HL. The element 99 in the first row and first column of HL is the product of the
first row matrix of H and the first column matrix of L:
CA MD 5
[12 13]c d Trick = 12152 + 13132 = 60 + 39 = 99
3 Slalom
Notice that $60 is the labor cost for assembling a trick ski at the California plant
and $39 is the labor cost for assembling a slalom ski at the Maryland plant.
Although both numbers represent labor costs, it makes no sense to add them to-
gether. They do not pertain to the same type of ski or to the same plant. So even
though the product HL happens to be defined mathematically, it has no useful inter-
pretation in this problem.
Now let’s consider the product LH. The element 70.5 in the first row and first
column of LH is given by the following product:
Assembly Finishing
12 Assembly
[5 1.5]c d = 51122 + 1.5172 = 60 + 10.5 = 70.5
7 Finishing
This time, $60 is the labor cost for assembling a trick ski at the California plant and
$10.50 is the labor cost for finishing a trick ski at the California plant. So the sum
is the total labor cost for producing a trick ski at the California plant. The other ele-
ments in LH also represent total labor costs, as indicated by the row and column
labels shown below:
Labor costs per ski

CA MD
$70.50 $77 Trick
LH = c d
$43 $47 Slalom
Figure 5 shows a solution to Example 9 on a spreadsheet.
Figure 5 Matrix multiplication in a spreadsheet: The command MMULT(B3:C4, F3:G4)

produces the matrix in B8:C9
Matched Problem 10 Refer to Example 10. The company wants to know how
many hours to schedule in each department in order to produce 2,000 trick skis and
1,000 slalom skis. These production requirements can be represented by either of
the following matrices:
Trick Slalom
skis skis
2,000 Trick skis
P = [2,000 1,000] Q = c d
1,000 Slalom skis
Using the labor-hour matrix L from Example 10, find PL or LQ, whichever has a mean-
ingful interpretation for this problem, and label the rows and columns accordingly.
CONCEPTUAL INSIGHT
Example 10 and Matched Problem 10 illustrate an important point about matrix
multiplication. Even if you are using a graphing calculator to perform the calcu-
lations in a matrix product, it is still necessary for you to know the definition of
matrix multiplication so that you can interpret the results correctly.
Exercises 3.4
Skills Warm-up Exercises In Problems 15–22, find the matrix product. Note that each
product can be found mentally, without the use of a calculator or
W In Problems 1–14, perform the indicated operation, if possible. (If pencil-and-paper calculations.
necessary, review the definitions at the beginning of Section 3.4.)
3 7 5 0 1 2 3 0 1 3
1. 31 54 + 33 104 2. c d - c d 15. c dc d 16. c dc d
2 -4 0 5 3 4 0 3 5 7
2 0 0 -4 -9 2 9 0 1 2 5 0 1 3 3 0
3. c d - c d 4. c d + c d 17. c dc d 18. c dc d
-3 6 -1 0 8 0 0 8 3 4 0 5 5 7 0 3
3 2 -6 1 0 1 3 5 0 0 2 4
5. c d + c - 1 9d 6. 4 c d 19. c dc d 20. c dc d
6 8 5 -3 0 0 7 9 1 0 6 8
4 3 5 0 1 2 4 0 0
7. 7 c 3 -5 9 4d 8. 310 124 + c d 21. c dc d 22. c dc d
3 7 9 0 0 6 8 1 0
3 4 -1 -1 1 4
A 9. c dc d 10. c dc d B Find the products in Problems 23–30.
-1 -2 2 2 -3 -2
2 -3 1 -1 -5 4 -6 7 -5 2
11. c dc d 12. c dc d 23. [3 - 2]c d 24. [- 3 5]c d
1 2 0 -2 3 -2 -8 9 4 -4
1 -1 2 -3 -6 7 -5 4 -5 2
13. c dc d 14. c dc d 25. c d [3 - 2] 26. c d [- 3 5]
0 -2 1 2 -8 9 3 -2 4 -4
2 6 0.6 0.4
54. A = [0.8 0.2] and B = c d
27. [- 1 0 1]C 1 S 28. [1 -2 3]C - 5 S 0.1 0.9
3 0
C 55. Find a, b, c, and d so that
2 6
29. C 1 S [- 1 0 1] 30. C - 5 S [1 -2 3] a b 2 -3 1 -2
c d + c d = c d
3 0 c d 0 1 3 -4
Problems 31–48 refer to the following matrices: 56. Find w, x, y, and z so that
2 -1 3 -3 1 7 -8 w x 4 -5
A = c d B = c d c d + c d = c d
0 4 -2 2 5 0 9 y z 6 0
-1 0 2 3 -2 57. Find a, b, c, and d so that

C = C 4 -3 1S D = C0 -1 S
1 -2 a b 1 0
-2 3 5 1 2 c dc d = c d
2 -3 c d 3 2
Perform the indicated operations, if possible.
58. Find a, b, c, and d so that
31. AC 32. CA
1 3 a b 6 -5
33. CD 34. DC c dc d = c d
1 4 c d 7 -7
35. B2 36. C 2
In Problems 59–62, determine whether the statement is true or
37. B + AD 38. C + DA
false.
39. 10.12DB 40. 10.22CD
59. There exist two 1 * 1 matrices A and B such that AB ∙ BA.
41. 132BA + 142AC 42. 122DB + 152CD
60. There exist two 2 * 2 matrices A and B such that AB ∙ BA.
43. 1 - 22BA + 162CD 44. 1 - 12AC + 132DB
61. There exist two nonzero 2 * 2 matrices A and B such that
45. ACD 46. CDA AB is the 2 * 2 zero matrix.
47. DBA 48. BAD 62. There exist two nonzero 1 * 1 matrices A and B such that
49. If a and b are nonzero real numbers, AB is the 1 * 1 zero matrix.
63. A square matrix is a diagonal matrix if all elements not on
a a a a
A = c d , and B = c d the principal diagonal are zero. So a 2 * 2 diagonal matrix
b b -a -a has the form
find AB and BA.
a 0
50. If a and b are nonzero real numbers, A = c d
0 d
a b a a where a and d are real numbers. Discuss the validity of each
A = c d, and B = c d
-a -b a a of the following statements. If the statement is always true,
find AB and BA. explain why. If not, give examples.
(A) If A and B are 2 * 2 diagonal matrices, then A + B is a
51. If a and b are nonzero real numbers and 2 * 2 diagonal matrix.
ab b2 (B) If A and B are 2 * 2 diagonal matrices, then AB is a
A = c d 2 * 2 diagonal matrix.
- a2 - ab
find A2. (C) If A and B are 2 * 2 diagonal matrices, then AB = BA.
64. A square matrix is an upper triangular matrix if all ele-
52. If a and b are nonzero real numbers and
ments below the principal diagonal are zero. So a 2 * 2
ab b - ab2 upper triangular matrix has the form
A = c d
a 1 - ab
a b
find A2. A = c d
0 d
In Problems 53 and 54, use a graphing calculator to calculate where a, b, and d are real numbers. Discuss the validity
B, B2, B3, c and AB, AB2, AB3, c . Describe any patterns you of each of the following statements. If the statement is
observe in each sequence of matrices. always true, explain why. If not, give examples.
0.4 0.6 (A) If A and B are 2 * 2 upper triangular matrices, then
53. A = [0.3 0.7] and B = c d A + B is a 2 * 2 upper triangular matrix.
0.2 0.8
(B) If A and B are 2 * 2 upper triangular matrices, then AB Labor-hours per shelving unit
is a 2 * 2 upper triangular matrix. Cutting Assembly Packaging
department department department
(C) If A and B are 2 * 2 upper triangular matrices, then
AB = BA. 0.7 hr 0.8 hr 0.3 hr Small shelving unit
M = C 1.1 hr 1.3 hr 0.4 hr S Medium shelving unit
1.6 hr 1.7 hr 0.5 hr Large shelving unit
Applications Hourly wages

IT PL
65. Cost analysis. A company with two different plants manu- ;12.45 ;9.95 Cutting department
factures guitars and banjos. Its production costs for each
N = C ;9.55 ;7.35 S Assembly department
instrument are given in the following matrices:
;7.75 ;5.65 Packaging department
Plant X Plant Y
Guitar Banjo Guitar Banjo (A) Find the labor costs for a medium shelving unit manu-
Materials $47 $39 $56 $42 factured at the Italian plant.
c d = A c d = B
Labor $90 $125 $84 $115
(B) Find the labor costs for a small shelving unit manufac-
Find 12 1A + B2, the average cost of production for the two tured at the Polish plant.
plants.
(C) Discuss possible interpretations of the elements in the
66. Cost analysis. If both labor and materials at plant X in Prob-
matrix products MN and NM.
lem 65 are increased by 20%, find 12 11.2A + B2, the new
average cost of production for the two plants. (D) If either of the products MN or NM has a meaningful inter-
pretation, find the product and label its rows and columns.
67. Markup. An import car dealer sells three models of a car.
The retail prices and the current dealer invoice prices (costs) 70. Inventory value. A personal computer retail company sells
for the basic models and options indicated are given in the five different computer models through three stores. The
following two matrices (where “Air” means air-conditioning): inventory of each model on hand in each store is summarized
Retail price in matrix M. Wholesale (W) and retail (R) values of each
model computer are summarized in matrix N.
Basic AM/FM Cruise
Car Air radio control Model
Model A $35,075 $2,560 $1,070 $640 A B C D E
Model B C $39,045 $1,840 $770 $460 S = M 4 2 3 7 1 Store 1
Model C $45,535 $3,400 $1,415 $850 M = C 2 3 5 0 6 S Store 2
10 4 3 4 3 Store 3
Dealer invoice price W R
Basic AM/FM Cruise $700 $840 A
Car Air radio control $1,400 $1,800 B
Model A $30,996 $2,050 $850 $510 N = E $1,800 $2,400 U C Model
Model B C $34,857 $1,585 $660 $395 S = N $2,700 $3,300 D
Model C $41,667 $2,890 $1,200 $725 $3,500 $4,900 E
We define the markup matrix to be M - N (markup is the (A) What is the retail value of the inventory at store 2?
difference between the retail price and the dealer invoice
price). Suppose that the value of the dollar has had a sharp (B) What is the wholesale value of the inventory at store 3?
decline and the dealer invoice price is to have an across-
(C) If either product MN or NM has a meaningful interpreta-
the-board 15% increase next year. To stay competitive with
tion, find the product and label its rows and columns.
domestic cars, the dealer increases the retail prices 10%.
What do the entries represent?
Calculate a markup matrix for next year’s models and the op-
tions indicated. (Compute results to the nearest dollar.) (D) Discuss methods of matrix multiplication that can be
68. Markup. Referring to Problem 67, what is the markup used to find the total inventory of each model on hand at
matrix resulting from a 20% increase in dealer invoice prices all three stores. State the matrices that can be used and
and an increase in retail prices of 15%? (Compute results to perform the necessary operations.
the nearest dollar.) 71. Cereal. A nutritionist for a cereal company blends two cere-
69. Labor costs. A company with manufacturing plants located als in three different mixes. The amounts of protein, carbohy-
in Italy (IT) and Poland (PL) has labor-hour and wage drate, and fat (in grams per ounce) in each cereal are given by
requirements for the manufacture of three types of shelving matrix M. The amounts of each cereal used in the three mixes
units as given in the following two matrices: are given by matrix N.
Cereal A Cereal B (C) If either product MN or NM has a meaningful interpreta-

4 g>oz 2 g>oz Protein tion, find the product and label its rows and columns.
What do the entries represent?
M = C 20 g>oz 16 g>oz S Carbohydrate
3 g>oz 1 g>oz Fat (D) Discuss methods of matrix multiplication that can be
used to find the total number of telephone calls, house
Mix X Mix Y Mix Z calls, and letters. State the matrices that can be used and
15 oz 10 oz 5 oz Cereal A perform the necessary operations.
N = c d
5 oz 10 oz 15 oz Cereal B 74. Test averages. A teacher has given four tests to a class of
(A) Find the amount of protein in mix X. five students and stored the results in the following matrix:
Tests
(B) Find the amount of fat in mix Z. 1 2 3 4
(C) Discuss possible interpretations of the elements in the Ann 78 84 81 86
matrix products MN and NM. Bob 91 65 84 92
(D) If either of the products MN or NM has a meaningful inter- Carol E 95 90 92 91 U = M
pretation, find the product and label its rows and columns. Dan 75 82 87 91
72. Heredity. Gregor Mendel (1822–1884) made discoveries Eric 83 88 81 76
that revolutionized the science of genetics. In one experi- Discuss methods of matrix multiplication that the teacher can use
ment, he crossed dihybrid yellow round peas (yellow and to obtain the information indicated below. In each case, state the
round are dominant characteristics; the peas also contained matrices to be used and then perform the necessary operations.
genes for the recessive characteristics green and wrinkled)
and obtained peas of the types indicated in the matrix: (A) The average on all four tests for each student, assuming
that all four tests are given equal weight
Round Wrinkled
Yellow 315 101 (B) The average on all four tests for each student, assuming
c d = M that the first three tests are given equal weight and the
Green 108 32
fourth is given twice this weight
Suppose he carried out a second experiment of the same type
and obtained peas of the types indicated in this matrix: (C) The class average on each of the four tests
Round Wrinkled Answers to Matched Problems

Yellow 370 128 a = -6
c d = N 1 5 13
Green 110 36 b = 7
1. C 0 -2 S 2. [- 1 -1 4] 3. 4. C 2 S
If the results of the two experiments are combined, discuss c = 9
2 1 35
matrix multiplication methods that can be used to find the d = -1
following quantities. State the matrices that can be used and
$235,000 $422,000
perform the necessary operations. 5. (A) c d
$372,000 $298,000
(A) The total number of peas in each category
$145,000 $268,000
(B) The total number of peas in all four categories (B) c d
$160,000 $254,000
(C) The percentage of peas in each category $9,500 $17,250
(C) c d
$13,300 $13,800
73. Politics. In a local London election, a public relations firm
promoted its candidate in three ways: telephone calls, house 9
6. [8] 7. [5 1.5]c d = [54], or $54
calls, and letters. The cost per contact is given in matrix M, 6
8. (A) Not defined
and the number of contacts of each type made in two adjacent
boroughs is given in matrix N. 2 2 -1 2
0 0
Cost per contact (B) C 1 6 12 -4 S (C) c d
0 0
-1 0 3 -2
£0.90 Telephone call
M = C £3.45 S House call -6 - 12
(D) c d (E) [11]
£1.15 Letter 3 6
12 -8 4
Telephone House
call call Letter (F) C 6 -4 2S
1,300 800 6,000 Greenwich 9 -6 3
N = c d
2,100 1200 7,000 Bromley 9. a = 4, c = 8, b = 9, d = - 2
(A) Find the total amount spent in Greenwich. Assembly Finishing
(B) Find the total amount spent in Bromley. 10. PL = [13,000 4,000] Labor@hours
3.5 Inverse of a Square Matrix

■■ Identity Matrix for Multiplication Identity Matrix for Multiplication
■■ Inverse of a Square Matrix Does the set of all matrices of a given size have an identity element for multiplica-
■■ Application: Cryptography tion? That is, if M is an arbitrary m * n matrix, does there exist an identity element
I such that IM = MI = M? The answer, in general, is no. However, the set of all
square matrices of order n (matrices with n rows and n columns) does have an
identity element.
DEFINITION Identity Matrix

The identity element for multiplication for the set of all square matrices of
order n is the square matrix of order n, denoted by I, with 1’s along the principal
diagonal (from the upper left corner to the lower right) and 0’s elsewhere.
For example,
1 0 0
1 0
c d and C0 1 0S
0 1
0 0 1
are the identity matrices for all square matrices of order 2 and 3, respectively.
Most graphing calculators have a built-in command for generating the identity
matrix of a given order (see Fig. 1).
Figure 1 Identity matrices
EXAMPLE 1 Identity Matrix Multiplication
1 0 0 3 -2 5 3 -2 5
(A) C 0 1 0S C 0 2 -3 S = C 0 2 -3 S
0 0 1 -1 4 -2 -1 4 -2
3 -2 5 1 0 0 3 -2 5
(B) C 0 2 -3 S C 0 1 0 S = C 0 2 -3 S
-1 4 -2 0 0 1 -1 4 -2
1 0 2 -1 3 2 -1 3
(C) c dc d = c d
0 1 -2 0 4 -2 0 4
1 0 0
2 -1 3 2 -1 3
(D) c d C0 1 0S = c d
-2 0 4 -2 0 4
0 0 1
SECTION 3.5 Inverse of a Square Matrix 203
Matched Problem 1 Multiply:

1 0 2 -3 2 -3 1 0
(A) c dc d and c dc d
0 1 5 7 5 7 0 1
1 0 0 4 2 4 2
1 0
(B) C 0 1 0S C3 -5 S and C3 -5 S c d
0 1
0 0 1 6 8 6 8
In general, we can show that if M is a square matrix of order n and I is the iden-
tity matrix of order n, then
IM = MI = M
If M is an m * n matrix that is not square 1 m ∙ n 2 , it is still possible to mul-
tiply M on the left and on the right by an identity matrix, but not with the same size
identity matrix (see Example 1C and D). To avoid the complications involved with
associating two different identity matrices with each nonsquare matrix, we restrict
our attention in this section to square matrices.
The only real number solutions to the equation x2 = 1 are x = 1 and x = - 1.

0 1
(A) Show that A = c d satisfies A2 = I, where I is the 2 * 2 identity.
1 0
0 -1
(B) Show that B = c d satisfies B2 = I.
-1 0
(C) Find a 2 * 2 matrix with all elements nonzero whose square is the 2 * 2 iden-
tity matrix.
Inverse of a Square Matrix

If r is an arbitary real number, then its additive inverse is the solution x to the equa-
tion r + x = 0. So the additive inverse of 3 is -3 , and the additive inverse of -7
is 7. Similarly, if M is an arbitary m * n matrix, then M has an additive inverse - M,
whose elements are just the additive inverses of the elements of M.
The situation is more complicated for multiplicative inverses. The multiplicative
inverse of an arbitrary real number r is the solution x to the equation r # x = 1.
1 - 15 - 4
So the multiplicative inverse of 3 is , and the multiplicative inverse of is .
3 4 15
Every real number has a multiplicative inverse except for 0. Because the equation
0 # x = 1 has no real solution, 0 does not have a multiplicative inverse.
Can we extend the multiplicative inverse concept to matrices? That is, given
a matrix M, can we find another matrix N such that MN = NM = I, the matrix
identity for multiplication? To begin, we consider the size of these matrices. Let M
be an n * m matrix and N a p * q matrix. If both MN and NM are defined, then
m = p and q = n (Fig. 2). If MN = NM, then n = p and q = m (Fig. 3). Thus,
we have m = p = n = q. In other words, M and N must be square matrices of the
same order. Later we will see that not all square matrices have inverses.
m5p q5n n5p
n3m p3q p3q n3m n3q p3m
M ? N 5 N ? M MN 5 NM
n3q p3m q5m
Figure 2 Figure 3
DEFINITION Inverse of a Square Matrix

Let M be a square matrix of order n and I be the identity matrix of order n. If there
exists a matrix M -1 (read “M inverse”) such that
M -1M = MM -1 = I
then M -1 is called the multiplicative inverse of M or, more simply, the inverse of
M. If no such matrix exists, then M is said to be a singular matrix.
Let us use the definition above to find M -1 for
2 3
M = c d
1 2
We are looking for
a c
M -1 = c d
b d
such that
MM -1 = M -1M = I
So we write
M M -1 I
2 3 a c 1 0
c dc d = c d
1 2 b d 0 1
and try to find a, b, c, and d so that the product of M and M -1 is the identity matrix I.
Multiplying M and M -1 on the left side, we obtain
12a + 3b2 12c + 3d2 1 0

c d = c d
1a + 2b2 1c + 2d2 0 1
which is true only if
2a + 3b = 1 2c + 3d = 0 Use Gauss–Jordan
elimination to solve
a + 2b = 0 c + 2d = 1 each system.
2 3 1 2 3 0
c ` d R1 4 R2 c ` d R1 4 R2
1 2 0 1 2 1
1 2 0 1 2 1
c ` d 1 - 22R1 + R2 S R2 c ` d 1 - 22R1 + R2 S R2
2 3 1 2 3 0
1 2 0 1 2 1
c ` d 1 - 12R2 S R2 c ` d 1 - 12R2 S R2
0 -1 1 0 -1 -2
1 2 0 1 2 1
c ` d 1 - 22R2 + R1 S R1 c ` d 1 - 22R2 + R1 S R1
0 1 -1 0 1 2
1 0 2 1 0 -3
c ` d c ` d
0 1 -1 0 1 2
a = 2, b = - 1 c = - 3, d = 2
a c 2 -3
M -1 = c d = c d
b d -1 2
CHECK
M M -1 I M -1 M
2 3 2 -3 1 0 2 -3 2 3
c dc d = c d = c dc d
1 2 -1 2 0 1 -1 2 1 2
Unlike nonzero real numbers, inverses do not always exist for square matrices.
For example, if
2 1
N = c d
4 2
then, using the previous process, we are led to the systems
2a + b = 1 2c + d = 0 Use Gauss–Jordan
elimination to solve
4a + 2b = 0 4c + 2d = 1 each system.
2 1 1 2 1 0
c ` d 1 - 22R1 + R2 S R2 c ` d 1 - 22R1 + R2 S R2
4 2 0 4 2 1
2 1 0 2 1 0
c ` d c ` d
0 0 -2 0 0 1
The last row of each augmented matrix contains a contradiction. So each system is
inconsistent and has no solution. We conclude that N -1 does not exist and N is a sin-
gular matrix.
Being able to find inverses, when they exist, leads to direct and simple solutions
to many practical problems. In the next section, we show how inverses can be used to
solve systems of linear equations.
The method outlined previously for finding M -1, if it exists, gets very involved
for matrices of order larger than 2. Now that we know what we are looking for, we
can use augmented matrices (see Sections 3.2 and 3.3) to make the process more
efficient.
EXAMPLE 2 Finding the Inverse of a Matrix Find the inverse, if it exists, of the matrix
1 -1 1
M = C0 2 -1 S
2 3 0
SOLUTION We start as before and write
M M -1 I
1 -1 1 a d g 1 0 0
C0 2 -1 S C b e hS = C0 1 0S
2 3 0 c f i 0 0 1
which is true only if
a - b + c = 1 d - e + f = 0 g - h + i = 0
2b - c = 0 2e - f = 1 2h - i = 0
2a + 3b = 0 2d + 3e = 0 2g + 3h = 1
Now we write augmented matrices for each of the three systems:

First Second Third
1 -1 1 1 1 -1 1 0 1 -1 1 0
C0 2 -1 3 0S C0 2 -1 3 1S C0 2 -1 † 0S
2 3 0 0 2 3 0 0 2 3 0 1
Since each matrix to the left of the vertical bar is the same, exactly the same row
operations can be used on each augmented matrix to transform it into a reduced
form. We can speed up the process substantially by combining all three augmented
matrices into the single augmented matrix form below:
1 -1 1 1 0 0
C0 2 -1 3 0 1 0 S = [M ∙ I ] (1)
2 3 0 0 0 1
We now try to perform row operations on matrix (1) until we obtain a row-equiva-
lent matrix of the form
I B
1 0 0 a d g
C0 1 0 3 b e hS = [I ∙ B] (2)
0 0 1 c f i
If this can be done, the new matrix B to the right of the vertical bar will be M -1.
Now let’s try to transform matrix (1) into a form like matrix (2). We follow the
same sequence of steps as we did in the solution of linear systems by Gauss–Jordan
elimination (see Section 3.3).
M I
1 -1 1 1 0 0
C0 2 -1 3 0 1 0S 1 - 22R1 + R3 S R3
2 3 0 0 0 1
1 -1 1 1 0 0
∼C 0 2 -1 3 0 1 0S 1
2 R2 S R2
0 5 -2 -2 0 1
1 -1 1 1 0 0
∼ C 0 1 - 12 3 0 1
0S R2 + R1 S R1
2
0 5 -2 -2 0 1 1 - 52R2 + R3 S R3
1 1
1 0 2 1 2 0
C
∼ 0 1 - 12 3 0 1
2 0S 2R3 S R3
0 0 1
2 -2 - 52 1
1 1
1 0 2 1 2 0 1 - 12 2R3 + R1 S R1
C
∼ 0 1 - 12 3 0 1
2 0S 1
2 R3 + R2 S R2
0 0 1 -4 -5 2
1 0 0 3 3 -1
∼C 0 1 0 3 -2 -2 1 S = [I ∙ B]
0 0 1 -4 -5 2
Converting back to systems of equations equivalent to our three original

systems, we have
a = 3 d = 3 g = -1
b = -2 e = -2 h = 1
c = -4 f = -5 i = 2
-1
And these are just the elements of M that we are looking for!
3 3 -1
-1
M = C -2 -2 1S
-4 -5 2
Note that this is the matrix to the right of the vertical line in the last augmented
matrix. That is, M -1 = B.
Since the definition of matrix inverse requires that
M -1 M = I and MM -1 = I (3)
it appears that we must compute both M -1 M and MM -1 to check our work. How-
ever, it can be shown that if one of the equations in (3) is satisfied, the other is also
satisfied. So to check our answer it is sufficient to compute either M -1 M or MM -1;
we do not need to do both.
CHECK
3 3 -1 1 -1 1 1 0 0
-1
M M = C -2 -2 1S C0 2 -1 S = C 0 1 0S = I
-4 -5 2 2 3 0 0 0 1
3 -1 1
Matched Problem 2 Let M = C - 1 1 0S.
1 0 1
(A) Form the augmented matrix [M ∙ I].

(B) Use row operations to transform [M ∙ I] into [I ∙ B].
(C) Verify by multiplication that B = M -1 (that is, show that BM = I).
The procedure shown in Example 2 can be used to find the inverse of any square
matrix, if the inverse exists, and will also indicate when the inverse does not exist.
These ideas are summarized in Theorem 1.
THEOREM 1 Inverse of a Square Matrix M

If [M ∙ I] is transformed by row operations into [I ∙ B], then the resulting matrix B
is M -1. However, if we obtain all 0’s in one or more rows to the left of the vertical
line, then M - 1 does not exist.

(A) Suppose that the square matrix M has a row of all zeros. Explain why M has no
inverse.
(B) Suppose that the square matrix M has a column of all zeros. Explain why M has
no inverse.
4 -1
EXAMPLE 3 Finding a Matrix Inverse Find M -1, given M = c d.
-6 2
SOLUTION 4 -1 2 1 0 1
c d 4 R1 S R1
-6 2 0 1
1 - 14 1
0
~c 2 4
d 6R1 + R2 S R2
-6 2 0 1
1 - 14 1
0
~c 1
2 4
3 d 2R2 S R2
0 2 2 1
1 - 14 1
0
~c 2 4
d 1
4 R2 + R1 S R1
0 1 3 2
1
1 0 1
~c 2 2
d
0 1 3 2
Therefore,
1
1 2
M -1 = c d
3 2
Check by showing that M - 1 M = I.
2 -6
Matched Problem 3 Find M -1, given M = c d.
1 -2
Most graphing calculators and spreadsheets can compute matrix inverses, as il-
lustrated in Figure 4 for the solution to Example 3.
(A) The command [A]21 produces the (B) The command MINVERSE (B2:C3)
inverse on this graphing calculator produces the inverse in this spreadsheet
Figure 4 Finding a matrix inverse

The inverse of
a b
A = c d
c d
is
d -b
ad - bc ad - bc 1 d -b
A- 1 = D T = c d D = ad - bc
-c a D -c a
ad - bc ad - bc
provided that D ∙ 0.
(A) Use matrix multiplication to verify this formula. What can you conclude about
A - 1 if D = 0?
(B) Use this formula to find the inverse of matrix M in Example 3.
2 -4
EXAMPLE 4 Finding a Matrix Inverse Find M -1, given M = c d.
-3 6
SOLUTION 2 -4 2 1 0 1
c d 2 R1 S R1
-3 6 0 1
1
1 -2 2 2 0
∼c d 3R1 + R2 S R2
-3 6 0 1
1
1 -2 0
∼c 2 2
3 d
0 0 2 1
We have all 0’s in the second row to the left of the vertical bar; therefore, the inverse
does not exist.
3 1
Matched Problem 4 Find N - 1, given N = c d.
6 2
Square matrices that do not have inverses are called singular matrices.
Graphing calculators and spreadsheets recognize singular matrices and generally
respond with some type of error message, as illustrated in Figure 5 for the solution
to Example 4.
(A) A graphing calculator displays (B) A spreadsheet displays a more

a clear error message cryptic error message
Figure 5
Application: Cryptography
Matrix inverses can provide a simple and effective procedure for encoding and
decoding messages. To begin, assign the numbers 1–26 to the letters in the alphabet,
as shown below. Also assign the number 0 to a blank to provide for space between
words. (A more sophisticated code could include both capital and lowercase letters and
punctuation symbols.)
Blank A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
The message “SECRET CODE” corresponds to the sequence
19 5 3 18 5 20 0 3 15 4 5
Any matrix whose elements are positive integers and whose inverse exists can be
used as an encoding matrix. For example, to use the 2 * 2 matrix
4 3
A = c
d
1 1
to encode the preceding message, first we divide the numbers in the sequence into
groups of 2 and use these groups as the columns of a matrix B with 2 rows:
19 3 5 0 15 5 Proceed down the columns,
B = c d not across the rows.
5 18 20 3 4 0
Notice that we added an extra blank at the end of the message to make the columns
come out even. Then we multiply this matrix on the left by A:
4 3 19 3 5 0 15 5
AB = c dc d
1 1 5 18 20 3 4 0
91 66 80 9 72 20
= c d
24 21 25 3 19 5
The coded message is
91 24 66 21 80 25 9 3 72 19 20 5
This message can be decoded simply by putting it back into matrix form and multi-
plying on the left by the decoding matrix A - 1. Since A - 1 is easily determined if A
is known, the encoding matrix A is the only key needed to decode messages that are
encoded in this manner.
EXAMPLE 5 Cryptography The message
46 84 85 28 47 46 4 5 10 30 48 72 29 57 38 38 57 95
was encoded with the matrix A shown next. Decode this message.
1 1 1
A = C2 1 2S
2 3 1
SOLUTION Since the encoding matrix A is 3 * 3, we begin by entering the coded

message in the columns of a matrix C with three rows:
46 28 4 30 29 38
C = C 84 47 5 48 57 57 S
85 46 10 72 38 95
If B is the matrix containing the uncoded message, then B and C are related by
C = AB. To recover B, we find A - 1 (details omitted) and multiply both sides of the
equation C = AB by A - 1:
B = A - 1C
-5 2 1 46 28 4 30 29 38
= C 2 -1 0 S C 84 47 5 48 57 57 S
4 -1 -1 85 46 10 72 38 95
23 0 0 18 7 19
= C 8 9 3 12 1 19 S
15 19 1 0 21 0
Writing the numbers in the columns of this matrix in sequence and using the corre-
spondence between numbers and letters noted earlier produces the decoded message:
23 8 15 0 9 19 0 3 1 18 12 0 7 1 21 19 19 0
W H O I S C A R L G A U S S
The answer to this question can be found earlier in this chapter.
Matched Problem 5 The message below was also encoded with the matrix A
in Example 5. Decode this message:
46 84 85 28 47 46 32 41 78 25 42 53 25 37 63 43 71 83 19 37 25
Exercises 3.5
Skills Warm-up Exercises 0 0 2 -3 2-3 0 0
11. (A) c dc d (B) c dc d
W In Problems 1–4, find the additive inverse and the multiplicative 0 1 4 5 4 5 0 1
inverse, if defined, of each real number. (If necessary, review 0 0 6 -7 6 -7 0 0
Section B.1). 0 1 -8 9 -8 9 0 1
1. (A) 4 (B) - 3 (C) 0 1 0 2 -3 2 -3 1 0
2. (A) - 7 (B) 2 (C) - 1 0 1 4 5 4 5 0 1
2 -1 1 0 -1 6 -1 6 1 0
3. (A) (B) (C) 1.6 14. (A) c dc d (B) c dc d
3 7 0 1 5 2 5 2 0 1
6 23
4. (A) (B) (C) - 0.8 1 0 0 -2 1 3
7 45
A In Problems 5–8, does the given matrix have a multiplicative 15. C 0 1 0S C 2 4 -2 S
inverse? Explain your answer. 0 0 1 5 1 0
4 0 0 1 0 1 0 0 1 -2 3
5. c 2 5 d 6. c d 7. c d 8. c d
8 0 0 0 1 16. C 0 1 0 S C -4 5 -6 S
In Problems 9–18, find the matrix products. Note that each 0 0 1 7 -8 9
product can be found mentally, without the use of a calculator or -2 1 3 1 0 0
pencil-and-paper calculations. 17. C 2 4 -2 S C 0 1 0S
1 0 2 -3 2 -3 1 0 5 1 0 0 0 1
0 0 4 5 4 5 0 0 1 -2 3 1 0 0
1 0 -1 6 -1 6 1 0 18. C - 4 5 -6 S C 0 1 0S
0 0 5 2 5 2 0 0 7 -8 9 0 0 1
In Problems 19–28, examine the product of the two matrices to 1 -3 0 2 3 0

determine if each is the inverse of the other. 45. C 0 1 1S 46. C 1 2 3S
3 -4 3 4 2 -1 4 0 -1 -5
19. c d; c d
-2 3 2 3
1 1 0 1 1 1
-2 -1 1 -1 47. C 2 3 -1 S 48. C 2 1 0S
20. c d; c d
-4 2 2 -2 1 0 2 4 3 1
2 2 1 1
21. c d; c d
-1 -1 -1 -1 Find the inverse of each matrix in Problems 49–56, if it exists.
5 -7 3 7
22. c d; c d 4 3 -4 3
-2 3 2 5 49. c d 50. c d
-3 -2 -5 4
-5 2 3 -2
23. c d; c d 2 6 2 -4
-8 3 8 -5 51. c d 52. c d
3 9 -3 6
7 4 3 4
24. c d; c d 2 1 -5 3
-5 -3 -5 -7 53. c d 54. c d
4 3 2 -2
1 -3 -5 -1 -1 0 3 -1 5 - 10
25. C - 2 3 5S; C 1 3 -5 S 55. c d 56. c d
-5 35 -2 24
-1 1 2 -1 -2 3
In Problems 57–60, find the inverse. Note that each inverse can
1 0 1 1 0 -1 be found mentally, without the use of a calculator or pencil-and-
26. C - 3 1 -2 S ; C3 1 -1 S paper calculations.
0 0 1 0 0 1
3 0 5 0
57. c d 58. c d
3 -1 1 1 -2 0 0 3 0 5
27. C 2 5 -4 S ; C 3 4 0S 2 0 3 0
0 2 1 -2 1 0 59. c 1d 60. c d
0 2 0 - 14
1 0 -1 1 0 -1 C Find the inverse of each matrix in Problems 61–68, if it exists.
28. C 3 1 -1 S ; C -3 1 -2 S
0 0 0 0 0 1 -5 -2 -2 2 -2 4
61. C 2 1 0S 62. C 1 1 1S
Without performing any row operations, explain why each of the 1 0 1 1 0 1
matrices in Problems 29–38 does not have an inverse.
2 1 1 1 -1 0
1 2 0 -2 3 -1 63. C 1 1 0S 64. C 2 -1 1S
29. c d 30. c d
-3 2 -1 4 0 1 -1 -1 0 0 1 1
1 -2 0 -1 -1 -2 2 4 2 2
31. C 3 0S 32. C 2 -2 S 65. C 4 3 0S 66. C 4 2 0S
2 -1 1 -3 4 0 4 5 0 5
0 1 0 -1 2 -1 -2 -1 -1 4
33. c d 34. c d
0 -2 0 2 67. C - 4 2 8S 68. C 3 3 - 22 S
1 -2 -1 2 6 -2 -1 -2 -1 19
35. c d 36. c d
0 0 0 0 4 3
69. Show that 1A-1 2 -1 = A for: A = c d
1 -3 -1 2 3 2
37. c d 38. c d
-3 9 1 -2 70. Show that 1AB2 -1 = B -1A-1 for
B Given M in Problems 39–48, find M -1 and show that M -1M = I.
4 3 2 5
A = c d and B = c d
-1 0 1 -5 3 2 3 7
39. c d 40. c d
-3 1 0 -1
71. Discuss the existence of M -1 for 2 * 2 diagonal matrices of
1 2 2 1 the form
41. c d 42. c d
1 3 5 3
a 0
M = cd
1 2 2 1 0 d
43. c d 44. c d
4 9 1 1 Generalize your conclusions to n * n diagonal matrices.
72. Discuss the existence of M -1 for 2 * 2 upper triangular 84. Cryptography. The following message was encoded with
matrices of the form matrix B. Decode this message:
75 61 28 94 35 22 13 40 49 21 16 52
a b
M = c d 42 45 19 64 38 55 10 65 69 75 24 102
0 d 67 49 19 82 10 5 5 10
Generalize your conclusions to n * n upper triangular Problems 85–88 require the use of a graphing calculator or a
matrices. computer. Use the 5 * 5 encoding matrix C given below. Form a
In Problems 73–75, find A-1 and A2. matrix with 5 rows and as many columns as necessary to accom-
modate the message.
-1 3 -5 4 5 -3
73. c d 74. c d 75. c d
0 1 -6 5 8 -5 1 0 1 0 1
0 1 1 0 3
76. Based on your observations in Problems 73–75, if A = A-1
for a square matrix A, what is A2? Give a mathematical argu- C = E2 1 1 1 1U
ment to support your conclusion. 0 0 1 0 2
1 1 1 2 1
85. Cryptography. Encode the message “THE EAGLE HAS

Applications LANDED” using matrix C.
86. Cryptography. Encode the message “ONE IF BY LAND
Problems 77–80 refer to the encoding matrix
AND TWO IF BY SEA” using matrix C.
1 2 87. Cryptography. The following message was encoded with

A = c d matrix C. Decode this message:
1 3
37 72 58 45 56 30 67 50 46 60 27 77
77. Cryptography. Encode the message “WINGARDIUM
41 45 39 28 24 52 14 37 32 58 70 36
LEVIOSA” using matrix A.
76 22 38 70 12 67
78. Cryptography. Encode the message “FINITE
88. Cryptography. The following message was encoded with
INCANTATEM” using matrix A.
matrix C. Decode this message:
79. Cryptography. The following message was encoded with
25 75 55 35 50 43 83 54 60 53 25 13
matrix A. Decode this message:
59 9 53 15 35 40 15 45 33 60 60 36
52 70 17 21 5 5 29 43 4 4 52 70 25 51 15 7 37 0 22
35 29 33 15 18 5 5
80. Cryptography. The following message was encoded with Answers to Matched Problems
matrix A. Decode this message:
4 2
36 44 5 5 38 56 55 75 18 23 56 75 2 -3
1. (A) c d (B) C 3 -5 S
22 33 37 55 27 40 53 79 59 81 5 7
6 8
Problems 81–84 require the use of a graphing calculator or
computer. Use the 4 * 4 encoding matrix B given below. Form a 3 -1 1 1 0 0
matrix with 4 rows and as many columns as necessary to accom- 2. (A) C - 1 1 0 3 0 1 0S
modate the message. 1 0 1 0 0 1
1 0 0 1 1 -1
2 2 1 3
(B) C 0 1 0 3 1 2 -1 S
1 2 2 1
B = E U 0 0 1 -1 -1 2
1 1 0 1
2 3 2 3 1 1 -1 3 -1 1 1 0 0
(C) C 1 2 -1 S C -1 1 0S = C0 1 0S
81. Cryptography. Encode the message “DEPART ISTANBUL -1 -1 2 1 0 1 0 0 1
ORIENT EXPRESS” using matrix B.
-1 3
82. Cryptography. Encode the message “SAIL FROM LISBON 3. c d
- 12 1
IN MORNING” using matrix B.
83. Cryptography. The following message was encoded with 4. Does not exist
matrix B. Decode this message: 5. WHO IS WILHELM JORDAN
85 74 27 109 31 27 13 40 139 73 58 154
61 70 18 93 69 59 23 87 18 13 9 22
3.6 Matrix Equations and Systems of Linear Equations

■■ Matrix Equations The identity matrix and inverse matrix discussed in the preceding section can be
■■ Matrix Equations and Systems
put to immediate use in the solution of certain simple matrix equations. Being able
of Linear Equations to solve a matrix equation gives us another important method of solving systems
of equations, provided that the system is independent and has the same number of
■■ Application variables as equations. If the system is dependent or if it has either fewer or more
variables than equations, we must return to the Gauss–Jordan method of elimination.
Matrix Equations
Solving simple matrix equations is similar to solving real number equations but with
two important differences:
1. there is no operation of division for matrices, and
2. matrix multiplication is not commutative.
Compare the real number equation 4x = 9 and the matrix equation AX = B.
The real number equation can be solved by dividing both sides of the equation by 4.
However, that approach cannot be used for AX = B, because there is no operation of
division for matrices. Instead, we note that 4x = 9 can be solved by multiplying both
sides of the equation by 14, the multiplicative inverse of 4. So we solve AX = B by
multiplying both sides of the equation, on the left, by A-1, the inverse of A. Because
matrix multiplication is not commutative, multiplying both sides of an equation on
the left by A-1 is different from multiplying both sides of an equation on the right by
A-1. In the case of AX = B, it is multiplication on the left that is required. The details
are presented in Example 1.
In solving matrix equations, we will be guided by the properties of matrices sum-
marized in Theorem 1.
THEOREM 1 Basic Properties of Matrices

Assuming that all products and sums are defined for the indicated matrices A, B, C,
I, and 0, then
Addition Properties
Associative: 1A + B2 + C = A + 1B + C2
Commutative: A + B = B + A
Additive identity: A + 0 = 0 + A = A
Additive inverse: A + 1 - A2 = 1 - A2 + A = 0
Multiplication Properties
Associative property: A1BC2 = 1AB2C
Multiplicative identity: AI = IA = A
Multiplicative inverse: If A is a square matrix and A-1 exists,
then AA-1 = A-1A = I.
Combined Properties
Left distributive: A1B + C2 = AB + AC
Right distributive: 1B + C2A = BA + CA
Equality
Addition: If A = B, then A + C = B + C.
Left multiplication: If A = B, then CA = CB.
Right multiplication: If A = B, then AC = BC.
SECTION 3.6 Matrix Equations and Systems of Linear Equations 215
EXAMPLE 1 Solving a Matrix Equation Given an n * n matrix A and n * 1 column matrices

B and X, solve AX = B for X. Assume that all necessary inverses exist.
SOLUTION We are interested in finding a column matrix X that satisfies the matrix
equation AX = B. To solve this equation, we multiply both sides on the left by A-1
to isolate X on the left side.
AX = B Use the left multiplication property.
-1 -1
A 1AX2 = A B Use the associative property.
-1 -1
1A A2X = A B A-1A = I
IX = A-1B IX = X
-1
X = A B
Matched Problem 1 Given an n * n matrix A and n * 1 column matrices B, C,

and X, solve AX + C = B for X. Assume that all necessary inverses exist.
! CAUTION Do not mix the left multiplication property and the right multipli-
cation property. If AX = B, then
A-1 1AX2 ∙ BA-1
Matrix Equations and Systems of Linear Equations

Now we show how independent systems of linear equations with the same number of
variables as equations can be solved. First, convert the system into a matrix equation
of the form AX = B, and then use X = A-1B as obtained in Example 1.
EXAMPLE 2 Using Inverses to Solve Systems of Equations Use matrix inverse methods to
solve the system:
x1 - x2 + x3 = 1
2x2 - x3 = 1 (1)
2x1 + 3x2 = 1
SOLUTION The inverse of the coefficient matrix
1 -1 1
A = C0 2 -1 S
2 3 0
provides an efficient method for solving this system. To see how, we convert system
(1) into a matrix equation:
A X B
1 -1 1 x1 1
C0 2 - 1 S C x2 S = C 1 S (2)
2 3 0 x3 1
Check that matrix equation (2) is equivalent to system (1) by finding the product
of the left side and then equating corresponding elements on the left with those on
the right.
We are interested in finding a column matrix X that satisfies the matrix equation
AX = B. In Example 1 we found that if A-1 exists, then
X = A-1B
The inverse of A was found in Example 2, Section 3.5, to be

3 3 -1
-1
A = C -2 -2 1S
-4 -5 2
Therefore,
X A-1 B
x1 3 3 -1 1 5
C x2 S = C - 2 -2 1 S C 1 S = C -3 S
x3 -4 -5 2 1 -7
and we can conclude that x1 = 5, x2 = - 3, and x3 = - 7. Check this result in
system (1).
Matched Problem 2 Use matrix inverse methods to solve the system:

3x1 - x2 + x3 = 1
- x1 + x2 = 3
x1 + x3 = 2
[Note: The inverse of the coefficient matrix was found in Matched Problem 2,
Section 3.5.]
At first glance, using matrix inverse methods seems to require the same amount
of effort as using Gauss–Jordan elimination. In either case, row operations must be
applied to an augmented matrix involving the coefficients of the system. The advan-
tage of the inverse matrix method becomes readily apparent when solving a number
of systems with a common coefficient matrix and different constant terms.
EXAMPLE 3 Using Inverses to Solve Systems of Equations Use matrix inverse methods to
solve each of the following systems:
(A) x1 - x2 + x3 = 3 (B) x1 - x2 + x3 = - 5
2x2 - x3 = 1 2x2 - x3 = 2
2x1 + 3x2 = 4 2x1 + 3x2 = -3
SOLUTION Notice that both systems have the same coefficient matrix A as system
(1) in Example 2. Only the constant terms have changed. We can use A -1 to solve
these systems just as we did in Example 2.
(A) X A -1 B
x1 3 3 -1 3 8
C x2 S = C - 2 -2 1 S C 1 S = C -4 S
x3 -4 -5 2 4 -9
x1 = 8, x2 = - 4, and x3 = - 9.
(B) X A -1 B
x1 3 3 -1 -5 -6
C x2 S = C - 2 -2 1S C 2S = C 3S
x3 -4 -5 2 -3 4
x1 = - 6, x2 = 3, and x3 = 4.
Matched Problem 3 Use matrix inverse methods to solve each of the follow-
ing systems (see Matched Problem 2):
(A) 3 x 1 - x 2 + x 3 = 3 (B) 3 x 1 - x 2 + x 3 = - 5
-x 1 + x 2 = -3 -x 1 + x 2 = 1
x1 + x3 = 2 x1 + x 3 = -4
As Examples 2 and 3 illustrate, inverse methods are very convenient for hand
calculations because once the inverse is found, it can be used to solve any new sys-
tem formed by changing only the constant terms. Since most graphing calculators
and computers can compute the inverse of a matrix, this method also adapts readily
to graphing calculator and spreadsheet solutions (Fig. 1). However, if your graphing
calculator (or spreadsheet) also has a built-in procedure for finding the reduced form
of an augmented matrix, it is just as convenient to use Gauss–Jordan elimination.
Furthermore, Gauss–Jordan elimination can be used in all cases and, as noted below,
matrix inverse methods cannot always be used.
Figure 1 Using inverse methods on a spreadsheet: The values in G2:G4 are produced by
the command MMULT (MINVERSE(B2:D4),F2:F4)
SUMMARY Using Inverse Methods to Solve Systems of Equations

If the number of equations in a system equals the number of variables and the
coefficient matrix has an inverse, then the system will always have a unique solu-
tion that can be found by using the inverse of the coefficient matrix to solve the
corresponding matrix equation.
Matrix equation Solution
AX = B X = A -1 B
CONCEPTUAL INSIGHT
There are two cases where inverse methods will not work:
Case 1. The coefficient matrix is singular.
Case 2. The number of variables is not the same as the number of equations.
In either case, use Gauss–Jordan elimination.
Application
The following application illustrates the usefulness of the inverse matrix method for
solving systems of equations.
EXAMPLE 4 Investment Analysis An investment advisor currently has two types of invest-
ments available for clients: a conservative investment A that pays 5% per year and
a higher risk investment B that pays 10% per year. Clients may divide their invest-
ments between the two to achieve any total return desired between 5% and 10%.
However, the higher the desired return, the higher the risk. How should each client
invest to achieve the indicated return?
Client
1 2 3 k
Total investment $20,000 $50,000 $10,000 k1
Annual return desired $ 1,200 $ 3,750 $ 900 k2
(6%) (7.5%) (9%)
SOLUTION The answer to this problem involves six quantities, two for each client.
Utilizing inverse matrices provides an efficient way to find these quantities. We will
solve the problem for an arbitrary client k with unspecified amounts k 1 for the total
investment and k 2 for the annual return. (Do not confuse k 1 and k 2 with variables.
Their values are known—they just differ for each client.)
Let x1 = amount invested in A by a given client
x2 = amount invested in B by a given client
Then we have the following mathematical model:
x1 + x2 = k1 Total invested
0 .0 5 x 1 + 0 .1 x 2 = k 2 Total annual return desired
Write as a matrix equation:
A X B
1 1 x k
c d c 1d = c 1d
0.05 0.1 x2 k2
If A-1 exists, then
X = A-1B
We now find A-1 by starting with the augmented matrix [A ∙ I] and proceeding as
discussed in Section 3.5:
1 1 1 0
c 2 d 20R2 S R2
0.05 0.1 0 1
1 1 1 0
∼c 2 d 1 - 12R1 + R2 S R2
1 2 0 20
1 1 1 0
∼c 2 d 1 - 12R2 + R1 S R1
0 1 -1 20
1 0 2 - 20
∼c 2 d
0 1 -1 20
Therefore,
A-1 A I
2 - 20 2 - 20 1 1 1 0
A-1 = c d Check: c dc d = c d
-1 20 -1 20 0.1 0.2 0 1
and
X A-1 B
x1 2 - 20 k1
c d = c dc d
x2 -1 20 k2
To solve each client’s investment problem, we replace k1 and k2 with appropri-
ate values from the table and multiply by A-1:
Client 1
x1 2 - 20 20,000 16,000
c d = c dc d = c d
x2 -1 20 1,200 4,000
Solution: x1 = $16,000 in investment A, x2 = $4,000 in investment B
Client 2
x1 2 - 20 50,000 25,000
c d = c dc d = c d
x2 -1 20 3,750 25,000
Client 3
x1 2 - 20 10,000 2,000
c d = c dc d = c d
x2 -1 20 900 8,000
Matched Problem 4 Repeat Example 4 with investment A paying 4% and

investment B paying 12%.
Figure 2 illustrates a solution to Example 4 on a spreadsheet.
Figure 2

Refer to the mathematical model in Example 4:
A X B
1 1 x k
c d c 1d = c 1d (3)
0.05 0.1 x2 k2
(A) Does matrix equation (3) always have a solution for any constant matrix B?
(B) Do all these solutions make sense for the original problem? If not, give
examples.
(C) If the total investment is k1 = $10,000, describe all possible annual returns k2.
Exercises 3.6
Skills Warm-up Exercises A Write Problems 9–12 as systems of linear equations without
matrices.
W In Problems 1–8, solve each equation for x, where x represents a
real number. (If necessary, review Section A.1). 3 1 x1 5
9. c dc d = c d
2 - 1 x2 -4
1. 5x = - 3 2. 4x = 9
-3 2 x1 1
3. 4x = 8x + 7 4. 6x = - 3x + 14 10. c dc d = c d
6 - 7 x2 -8
5. 6x + 8 = - 2x + 17 6. - 4x + 3 = 5x + 12
-3 1 0 x1 3
7. 10 - 3x = 7x + 9 8. 2x + 7x + 1 = 8x + 3 - x 11. C 2 0 1 S C x2 S = C - 4 S
-1 3 -2 x3 2
2 -1 0 x1 6 33. x1 + 2x2 = k1 34. 2x1 + x2 = k1

12. C - 2 3 - 1 S C x2 S = C - 4 S 4x1 + 9x2 = k2 x1 + x2 = k2
4 0 3 x3 7
(A) k1 = 1, k2 = 2 (A) k1 = - 1, k2 = - 2
(B) k1 = - 3, k2 = 4 (B) k1 = 2, k2 = 3
Write each system in Problems 13–16 as a matrix equation of the
form AX = B. (C) k1 = 2, k2 = - 1 (C) k1 = 2, k2 = 0
13. 3x1 - 4x2 = 1 14. 2x1 + x2 = 8 35. x1 - 3x2 = k1
2x1 + x2 = 5 - 5x1 + 3x2 = - 4 x2 + x3 = k2
15. x1 - 3x2 + 2x3 = - 3 16. 4x1 - 3x2 + 2x3 = - 1 2x1 - x2 + 4x3 = k3
- 2x1 + 3x2 = 1 - 3x1 + 5x2 = 7 (A) k1 = 1, k2 = 0, k3 = 2
x1 + x2 + 4x3 = - 2 2x1 - 6x3 = - 8 (B) k1 = - 1, k2 = 1, k3 = 0
Find x1 and x2 in Problems 17–20. (C) k1 = 2, k2 = - 2, k3 = 1
x1 3 -2 -2 x1 -2 1 3 36. 2x1 + 3x2 = k1
17. c d = c dc d 18. c d = c dc d
x2 1 4 1 x2 -1 2 -2
x1 + 2x2 + 3x3 = k2
x -4 3 2 x 3 -1 -2
19. c 1 d = c dc d 20. c 1 d = c dc d - x2 - 5x3 = k3
x2 5 -6 -7 x2 0 2 1
(A) k1 = 0, k2 = 2, k3 = 1
In Problems 21–24, find x1 and x2.
(B) k1 = - 2, k2 = 0, k3 = 1
1 - 1 x1 5 1 3 x1 9
21. c dc d = c d 22. c dc d = c d (C) k1 = 3, k2 = 1, k3 = 0
1 - 2 x2 7 1 4 x2 6
1 1 x1 15 1 1 x1 10 37. x1 + x2 = k1
23. c dc d = c d 24. c dc d = c d
2 - 3 x2 10 3 - 2 x2 20 2x1 + 3x2 - x3 = k2
In Problems 25–30, solve for x1 and x2. x1 + 2x3 = k3

1 3 x1 5 14 (A) k1 = 2, k2 = 0, k3 = 4
25. c dc d + c d = c d
1 1 x2 2 7 (B) k1 = 0, k2 = 4, k3 = - 2
2 1 x1 3 10 (C) k1 = 4, k2 = 2, k3 = 0
26. c dc d + c d = c d
1 3 x2 5 16
38. x1 + x2 + x3 = k1
2 6 x1 5 14 2x1 + x2 = k2
27. c dc d + c d = c d
-4 - 12 x2 2 7
4x1 + 3x2 + x3 = k3
2 1 x1 3 10
28. c dc d + c d = c d (A) k1 = 0, k2 = 3, k3 = 6
-6 - 3 x2 5 16
(B) k1 = - 3, k2 = 0, k3 = 6
4 2 x1 4 14
29. c dc d - c d = c d (C) k1 = 3, k2 = 6, k3 = 0
3 1 x2 3 8
3 2 x1 4 4 In Problems 39–44, the matrix equation is not solved correctly.
30. c dc d - c d = c d Explain the mistake and find the correct solution. Assume that
2 1 x2 1 3
the indicated inverses exist.
B In Problems 31–38, write each system as a matrix equation B B
and solve using inverses. [Note: The inverses were found in 39. AX = B, X = 40. XA = B, X =
A A
Problems 41–48, Exercises 3.5.]
41. XA = B, X = A-1B 42. AX = B, X = BA-1
31. x1 + 2x2 = k1 32. 2x1 + x2 = k1
43. AX = BA, X = A-1BA, X = B
x1 + 3x2 = k2 5x1 + 3x2 = k2
44. XA = AB, X = AB A-1, X = B
(A) k1 = 1, k2 = 3 (A) k1 = 2, k2 = 13
(B) k1 = 2, k2 = 4 In Problems 45–50, explain why the system cannot be solved by
(B) k1 = 3, k2 = 5
matrix inverse methods. Discuss methods that could be used and
(C) k1 = - 2, k2 = 1 (C) k1 = 1, k2 = - 3 then solve the system.
45. - 3x1 + 9x2 = - 6 46. - 3x1 + 9x2 = - 6
6x1 - 18x2 = 9 6x1 - 18x2 = 12
47. x1 - 3x2 - 2x3 = - 1 48. x1 - 3x2 - 2x3 = - 1 (A) How many tickets of each category should be sold to
bring in each of the returns indicated in the table?
- 2x1 + 6x2 + 4x3 = 3 - 2x1 + 7x2 + 3x3 = 3
(B) Is it possible to bring in a return of $200,000? Of
49. x1 - 2x2 + 3x3 = 1 50. x1 - 2x2 + 3x3 = 1
$400,000? Explain.
2x1 - 3x2 - 2x3 = 3 2x1 - 3x2 - 2x3 = 3
(C) Describe all the possible returns.
x1 - x2 - 5x3 = 2 x1 - x2 - 5x3 = 4
64. Parking receipts. Parking fees at a zoo are $5.00 for local
C For n * n matrices A and B, and n * 1 column matrices C, D, residents and $7.50 for all others. At the end of each day,
and X, solve each matrix equation in Problems 51–56 for X. the total number of vehicles parked that day and the gross
Assume that all necessary inverses exist. receipts for the day are recorded, but the number of vehicles
in each category is not. The following table contains the
51. AX - BX = C 52. AX + BX = C relevant information for a recent 4-day period:
53. AX + X = C 54. AX - X = C Day
55. AX - C = D - BX 56. AX + C = BX + D 1 2 3 4
In Problems 57 and 58, solve for x1 and x2. Vehicles parked 1,200 1,550 1,740 1,400
Gross receipts $7,125 $9,825 $11,100 $8,650
5 10 x1 2 - 1 x1 97
57. c dc d - c dc d = c d (A) How many vehicles in each category used the zoo’s
4 7 x2 3 3 x2 35
parking facilities each day?
5 1 x1 -3 2 x1 20
58. c dc d + c dc d = c d (B) If 1,200 vehicles are parked in one day, is it possible to
2 2 x2 1 3 x2 31
take in gross receipts of $5,000? Of $10,000? Explain.
In Problems 59–62, write each system as a matrix equation and
(C) Describe all possible gross receipts on a day when
solve using the inverse coefficient matrix. Use a graphing calcu-
1,200 vehicles are parked.
lator or computer to perform the necessary calculations.
65. Production scheduling. A supplier manufactures car and
59. x1 + 5x2 + 6x3 = 76 60. 8x1 + 7x2 + 6x3 = 25
truck frames at two different plants. The production rates
2x1 + 3x2 + 8x3 = 92 2x1 + 3x2 + 4x3 = 37 (in frames per hour) for each plant are given in the table:
11x1 + 9x2 + 4x3 = 181 x1 + x3 = 49 Plant Car Frames Truck Frames
61. 2x1 + x2 + 5x3 + 5x4 = 37 A 10 5

B 8 8
3x1 - 4x2 + 3x3 + 2x4 = 0
How many hours should each plant be scheduled to operate
7x1 + 3x2 + 8x3 + 4x4 = 45
to exactly fill each of the orders in the following table?
5x1 + 9x2 + 6x3 + 7x4 = 94
Orders
62. x1 + 4x2 - x3 + 4x4 = 11 1 2 3
3x1 + x2 + 4x3 + x4 = 27 Car frames 3,000 2,800 2,600
Truck frames 1,600 2,000 2,200
2x1 + 7x2 + x3 - 8x4 = 31
66. Production scheduling. Labor and material costs for manu-
6x1 + 4x3 - 5x4 = 47 facturing two guitar models are given in the table:
Guitar Model Labor Cost Material Cost
Applications A $30 $20

B $40 $30
Construct a mathematical model for each of the following prob-
lems. (The answers in the back of the book include both the math- (A) If a total of $3,000 a week is allowed for labor and
ematical model and the interpretation of its solution.) Use matrix material, how many of each model should be produced
inverse methods to solve the model and then interpret the solution. each week to use exactly each of the allocations of the
$3,000 indicated in the following table?
63. Concert tickets. A concert hall has 10,000 seats and two
categories of ticket prices, $25 and $35. Assume that all Weekly Allocation
seats in each category can be sold. 1 2 3
Labor $1,800 $1,750 $1,720
Concert
Material $1,200 $1,250 $1,280
1 2 3
Tickets sold 10,000 10,000 10,000 (B) Is it possible to use an allocation of $1,600 for labor
Return required $275,000 $300,000 $325,000 and $1,400 for material? Of $2,000 for labor and
$1,000 for material? Explain.
67. Incentive plan. A small company provides an incentive taught and the annual salary (in thousands of dollars) for
plan for its top executives. Each executive receives as a each rank are given in the table.
bonus a percentage of the portion of the annual profit that Rank
remains after the bonuses for the other executives have been Lecturer Instructor
deducted (see the table). If the company has an annual profit Sections taught 3 4
of $2 million, find the bonus for each executive. Round each
Annual salary (thousand $) 20 25
bonus to the nearest hundred dollars.
The number of sections taught by new faculty and the
Officer Bonus
amount budgeted for salaries (in thousands of dollars) at
President 3% each of the colleges are given in the following table. How
Executive vice president 2.5% many faculty of each rank should be hired at each college
Associate vice president 2% to exactly meet the demand for sections and completely
Assistant vice president 1.5% exhaust the salary budget?
68. Incentive plan. Repeat Problem 67 if the company decides Community College
to include a 0.7% bonus for the sales manager in the incen- 1 2 3
tive plan. Demand for sections 30 33 35
Salary budget (thousand $) 200 210 220
69. Diets. A biologist has available two commercial food
mixes containing the percentage of protein and fat given in
the table. Answers to Matched Problems
1. AX + C = B
Mix Protein(%) Fat(%)
1AX + C2 - C = B - C
A 20 4
B 14 3 AX + 1C - C2 = B - C
AX + 0 = B - C
(A) How many ounces of each mix should be used to pre-
AX = B - C
pare each of the diets listed in the following table?
A 1AX2 = A-1 1B - C2
-1
Diet 1A-1A2X = A-1 1B - C2

1 2 3
IX = A-1 1B - C2
Protein 80 oz 90 oz 100 oz
X = A-1 1B - C2
Fat 17 oz 18 oz 21 oz
2. x1 = 2, x2 = 5, x3 = 0
(B) Is it possible to prepare a diet consisting of 100 ounces 3. (A) x1 = - 2, x2 = - 5, x3 = 4
of protein and 22 ounces of fat? Of 80 ounces of protein (B) x1 = 0, x2 = 1, x3 = - 4
and 15 ounces of fat? Explain.
1.5 - 12.5
70. Education. A state university system is planning to hire 4. A-1 = c d ; client 1: $15,000 in A and $5,000
new faculty at the rank of lecturer or instructor for several - 0.5 12.5
in B; client 2: $28,125 in A and $21,875 in B; client 3: $3,750
of its two-year community colleges. The number of sections
in A and $6,250 in B
3.7 Leontief Input–Output Analysis

■■ Two-Industry Model An important application of matrices and their inverses is input–output analysis.
■■ Three-Industry Model
Wassily Leontief (1905–1999), the primary force behind this subject, was awarded
the Nobel Prize in economics in 1973 because of the significant impact his work had
on economic planning for industrialized countries. Among other things, he conducted
a comprehensive study of how 500 sectors of the U.S. economy interacted with each
other. Of course, large-scale computers played a crucial role in this analysis.
Our investigation will be more modest. In fact, we start with an economy com-
prised of only two industries. From these humble beginnings, ideas and definitions
will evolve that can be readily generalized for more realistic economies. Input–output
analysis attempts to establish equilibrium conditions under which industries in an
economy have just enough output to satisfy each other’s demands in addition to final
(outside) demands.
SECTION 3.7 Leontief Input–Output Analysis 223
Two-Industry Model
We start with an economy comprised of only two industries, electric company E and
water company W. Output for both companies is measured in dollars. The electric
company uses both electricity and water (inputs) in the production of electricity (out-
put), and the water company uses both electricity and water (inputs) in the production
of water (output). Suppose that the production of each dollar’s worth of electricity
requires $0.30 worth of electricity and $0.10 worth of water, and the production of
each dollar’s worth of water requires $0.20 worth of electricity and $0.40 worth of
water. If the final demand (the demand from all other users of electricity and water) is
d1 = $12 million for electricity
d2 = $8 million for water
how much electricity and water should be produced to meet this final demand?
To begin, suppose that the electric company produces $12 million worth of elec-
tricity and the water company produces $8 million worth of water. Then the produc-
tion processes of the companies would require
Electricity Electricity
required to required to
produce produce
electricity water
0.31122 + 0.2182 = $5.2 million of electricity
and
Water Water
required to required to
produce produce
electricity water
0.11122 + 0.4182 = $4.4 million of water
leaving only $6.8 million of electricity and $3.6 million of water to satisfy the final
demand. To meet the internal demands of both companies and to end up with enough
electricity for the final outside demand, both companies must produce more than just
the final demand. In fact, they must produce exactly enough to meet their own inter-
nal demands plus the final demand. To determine the total output that each company
must produce, we set up a system of equations.
If
x1 = total output from electric company
x2 = total output from water company
then, reasoning as before, the internal demands are
0.3x1 + 0.2x2 Internal demand for electricity
0.1x1 + 0.4x2 Internal demand for water
Combining the internal demand with the final demand produces the following system
of equations:
Total Internal Final
output demand demand
x1 = 0.3x1 + 0.2x2 + d1 (1)
x2 = 0.1x1 + 0.4x2 + d2
or, in matrix form,
x1 0.3 0.2 x1 d1
c d = c dc d + c d
x2 0.1 0.4 x2 d2
or
X = MX + D (2)
where
d1
D = c d Final demand matrix
d2
x1
X = c d Output matrix
x2
E W
E 0.3 0.2
M = c d Technology matrix
W 0.1 0.4
The technology matrix is the heart of input–output analysis. The elements in the
technology matrix are determined as follows (read from left to right and then up):
Output
E W
c c
input from E input from E
ES £ to produce $1 ≥ £ to produce $1 ≥
of electricity of water
Input G W = M
input from W input from W
W S £ to produce $1 ≥ £ to produce $1 ≥
of electricity of water
CONCEPTUAL INSIGHT
Labeling the rows and columns of the technology matrix with the first letter of
each industry is an important part of the process. The same order must be used
for columns as for rows, and that same order must be used for the entries of D
(the final demand matrix) and the entries of X (the output matrix). In this book we
normally label the rows and columns in alphabetical order.
Now we solve equation (2) for X. We proceed as in Section 3.6:

X = MX + D
X - MX = D
1 0
IX - MX = D I = c d
0 1
1I - M2X = D
X = 1I - M2 -1D Assuming I - M has an inverse (3)
Omitting the details of the calculations, we find
0.7 - 0.2 1.5 0.5

I - M = c d and 1I - M2 -1 = c d
- 0.1 0.6 0.25 1.75
Then we have
x1 1.5 0.5 d1 1.5 0.5 12 22

c d = c dc d = c dc d = c d (4)
x2 0.25 1.75 d2 0.25 1.75 8 17
Therefore, the electric company must produce an output of $22 million and the
water company must produce an output of $17 million so that each company can
meet both internal and final demands.
CHECK We use equation (2) to check our work:
X = MX + D
22 ≟ 0.3 0.2 22 12
c d c dc d + c d
17 0.1 0.4 17 8
22 ≟ 10 12
c d c d + c d
17 9 8
22 ✓ 22
c d = c d
17 17
To solve this input–output problem on a graphing calculator, simply store matri-
ces M, D, and I in memory; then use equation (3) to find X and equation (2) to check
your results. Figure 1 illustrates this process on a graphing calculator.
(A) Store M, D, and I in the graphing (B) Compute X and check in equation (2)
calculator's memory
Figure 1
Actually, equation (4) solves the original problem for arbitrary final demands d1
and d2. This is very useful, since equation (4) gives a quick solution not only for the
final demands stated in the original problem but also for various other projected final
demands. If we had solved system (1) by Gauss–Jordan elimination, then we would
have to start over for each new set of final demands.
Suppose that in the original problem the projected final demands 5 years from
now are d1 = 24 and d2 = 16. To determine each company’s output for this projec-
tion, we simply substitute these values into equation (4) and multiply:
x1 1.5 0.5 24 44
c d = c dc d = c d
x2 0.25 1.75 16 34
We summarize these results for convenient reference.
SUMMARY Solution to a Two-Industry Input–Output Problem

Given two industries, C1 and C2, with
Technology Output Final demand
matrix matrix matrix
C1 C2
C1 a11 a12 x1 d1
M = c d X = c d D = c d
C2 a21 a22 x2 d2
where aij is the input required from Ci to produce a dollar’s worth of output for Cj,
the solution to the input–output matrix equation
Total Internal Final
output demand demand
X = MX + D
is
X = 1I - M2 -1D (3)
assuming that I - M has an inverse.
Three-Industry Model
Equations (2) and (3) in the solution to a two-industry input–output problem are the
same for a three-industry economy, a four-industry economy, or an economy with n
industries (where n is any natural number). The steps we took going from equation
(2) to equation (3) hold for arbitrary matrices as long as the matrices have the correct
sizes and 1I - M2 -1 exists.

If equations (2) and (3) are valid for an economy with n industries, discuss the size of
all the matrices in each equation.
The next example illustrates the application of equations (2) and (3) to a three-
industry economy.
EXAMPLE 1 Input–Output Analysis An economy is based on three sectors, agriculture (A),

energy (E), and manufacturing (M). Production of a dollar’s worth of agriculture
requires an input of $0.20 from the agriculture sector and $0.40 from the energy
sector. Production of a dollar’s worth of energy requires an input of $0.20 from the
energy sector and $0.40 from the manufacturing sector. Production of a dollar’s
worth of manufacturing requires an input of $0.10 from the agriculture sector, $0.10
from the energy sector, and $0.30 from the manufacturing sector. Find the output
from each sector that is needed to satisfy a final demand of $20 billion for agricul-
ture, $10 billion for energy, and $30 billion for manufacturing.
SOLUTION Since this is a three-industry problem, the technology matrix will be
a 3 * 3 matrix, and the output and final demand matrices will be 3 * 1 column
matrices.
To begin, we form a blank 3 * 3 technology matrix and label the rows and
columns in alphabetical order.
Technology matrix
Output
A E M
A
Input E C S = M
M
Now we analyze the production information given in the problem, beginning with
agriculture.
“Production of a dollar’s worth of agriculture requires an input of $0.20
from the agriculture sector and $0.40 from the energy sector.”
We organize this information in a table and then insert it in the technology matrix.
Since manufacturing is not mentioned in the agriculture production information, the
input from manufacturing is $0.
Agriculture
Input Output
0.2 A E M
A ¡ A A
0.4 0.2
E ¡ A E £ 0.4 §
0
M¡ A M 0
“Production of a dollar’s worth of energy requires an input of $0.20 from

the energy sector and $0.40 from the manufacturing sector.”
Energy
Input Output A E M
0
A ¡ E A 0.2 0
0.2
E ¡0.4
E E £ 0.4 0.2 §
M ¡ E M 0 0.4
“Production of a dollar’s worth of manufacturing requires an input of $0.10
from the agriculture sector, $0.10 from the energy sector, and $0.30 from the
manufacturing sector.”
Manufacturing
Input Output A E M
0.1
A ¡ M A 0.2 0 0.1
0.1
E ¡ M E £ 0.4 0.2 0.1 §
0.3
M ¡ M M 0 0.4 0.3
Therefore,
Technology matrix Final demand Output

A E M matrix matrix
A 0.2 0 0.1 20 x1
M = E £ 0.4 0.2 0.1 § D = £ 10 § X = £ x2 §
M 0 0.4 0.3 30 x3
where M, X, and D satisfy the input–output equation X = MX + D. Since the so-

lution to this equation is X = 1I - M2 -1D, we must first find I - M and then
1I - M2 -1. Omitting the details of the calculations, we have
0.8 0 - 0.1
I - M = C - 0.4 0.8 - 0.1 S
0 - 0.4 0.7
and
1.3 0.1 0.2

-1
1I - M2 = C 0.7 1.4 0.3 S
0.4 0.8 1.6
So the output matrix X is given by
X 1I - M2 -1 D
x1 1.3 0.1 0.2 20 33
C x2 S = C 0.7 1.4 0.3 S C 10 S = C 37 S
x3 0.4 0.8 1.6 30 64
An output of $33 billion for agriculture, $37 billion for energy, and $64 billion for
manufacturing will meet the given final demands. You should check this result in
equation (2).
Figure 2 illustrates a spreadsheet solution for Example 1.
Figure 2 The command MMULT(MINVERSE(F3:H5), J3:J5) produces the output in L3:L5
Matched Problem 1 An economy is based on three sectors, coal, oil, and

transportation. Production of a dollar’s worth of coal requires an input of $0.20
from the coal sector and $0.40 from the transportation sector. Production of a dol-
lar’s worth of oil requires an input of $0.10 from the oil sector and $0.20 from the
transportation sector. Production of a dollar’s worth of transportation requires an
input of $0.40 from the coal sector, $0.20 from the oil sector, and $0.20 from the
transportation sector.
(A) Find the technology matrix M.
(B) Find 1I - M2 -1.
(C) Find the output from each sector that is needed to satisfy a final demand of
$30 billion for coal, $10 billion for oil, and $20 billion for transportation.
Exercises 3.7
Skills Warm-up Exercises B Problems 15–20 pertain to the following input–output model:
Assume that an economy is based on three industrial sectors:
W In Problems 1–8, solve each equation for x, where x represents agriculture (A), building (B), and energy (E). The technology
a real number. (If necessary, review Section A.1.) matrix M and final demand matrices (in billions of dollars) are
1. x = 3x + 6 2. x = 4x - 5 A B E
3. x = 0.9x + 10 4. x = 0.6x + 84 A 0.2 0.3 0.2
B C 0.1 0.2 0.1 S = M
5. x = 0.2x + 3.2 6. x = 0.3x + 4.2
E 0.1 0.3 0.1
7. x = 0.68x + 2.56 8. x = 0.89x + 10.23 10 15
A Problems 9–14 pertain to the following input–output model: D1 = C 15 S D2 = C 10 S
Assume that an economy is based on two industrial sectors, 20 5
agriculture (A) and energy (E). The technology matrix M
and final demand matrices (in billions of dollars) are 15. How much input from A, B, and E are required to produce a
dollar’s worth of output for B?
A E
A 0.4 0.2 16. How much of each of E’s output dollars is required as input
c d = M for each of the three sectors?
E 0.2 0.1
6 8 12 17. Find I - M.
D1 = c d D2 = c d D3 = c d
4 5 9 18. Find 1I - M2-1 .
9. How much input from A and E are required to produce a Show that 1I - M2-1 1I - M2 = I.
dollar’s worth of output for A? 19. Use 1I - M2-1 in Problem 18 to find the output for each
sector that is needed to satisfy the final demand D1.
10. How much input from A and E are required to produce a
dollar’s worth of output for E? 20. Repeat Problem 19 for D2.
-1
11. Find I - M and 1I - M2 . In Problems 21–26, find 1I - M2-1 and X.
12. Find the output for each sector that is needed to satisfy the
final demand D1. 0.1 0.2 15
21. M = c d; D = c d
0.9 0.8 45
13. Repeat Problem 12 for D2.
0.4 0.2 30
14. Repeat Problem 12 for D3. 22. M = c d; D = c d
0.6 0.8 50
0.2 0.4 25 The management of these two sectors would like to set the
23. M = c d; D = c d total output level so that the final demand is always 40% of
0.4 0.3 35
the total output. Discuss methods that could be used to ac-
0.4 0.1 15 complish this objective.
24. M = c d; D = c d
0.2 0.3 20 32. The technology matrix for an economy based on automobiles
0.3 0.1 0.3 20 (A) and construction (C) is
C 25. M = C 0.2 0.1 0.2 S ; D = C 5 S A C
0.1 0.1 0.1 10 A 0.1 0.4
M = c d
C 0.1 0.1
0.3 0.2 0.3 10
26. M = C 0.1 0.1 0.1 S ; D = C 25 S The management of these two sectors would like to set the
0.1 0.2 0.1 15 total output level so that the final demand is always 70% of
the total output. Discuss methods that could be used to ac-
complish this objective.
27. The technology matrix for an economy based on agriculture
(A) and manufacturing (M) is 33. All the technology matrices in the text have elements between
A M 0 and 1. Why is this the case? Would you ever expect to find
A 0.3 0.25 an element in a technology matrix that is negative? That is
M = c d equal to 0? That is equal to 1? That is greater than 1?
M 0.1 0.25
(A) Find the output for each sector that is needed to satisfy a 34. The sum of the elements in a column of any of the technol-
final demand of $40 million for agriculture and $40 mil- ogy matrices in the text is less than 1. Why is this the case?
lion for manufacturing. Would you ever expect to find a column with a sum equal
to 1? Greater than 1? How would you describe an economic
(B) Discuss the effect on the final demand if the agriculture system where the sum of the elements in every column of the
output in part (A) is increased by $20 million and manu- technology matrix is 1?
facturing output remains unchanged.
28. The technology matrix for an economy based on energy (E)
and transportation (T) is
E T Applications
E 0.25 0.25
M = c d
T 0.4 0.2 35. Coal, steel. An economy is based on two industrial sectors,
coal and steel. Production of a dollar’s worth of coal requires
(A) Find the output for each sector that is needed to satisfy a
an input of $0.10 from the coal sector and $0.20 from the
final demand of $50 million for energy and $50 million
steel sector. Production of a dollar’s worth of steel requires
for transportation.
an input of $0.20 from the coal sector and $0.40 from the
(B) Discuss the effect on the final demand if the transporta- steel sector. Find the output for each sector that is needed to
tion output in part (A) is increased by $40 million and satisfy a final demand of $20 billion for coal and $10 billion
the energy output remains unchanged. for steel.
29. Refer to Problem 27. Fill in the elements in the following 36. Transportation, manufacturing. An economy is based on
technology matrix. two sectors, transportation and manufacturing. Production
M A of a dollar’s worth of transportation requires an input of
M $0.10 from each sector and production of a dollar’s worth of
T = c d
A manufacturing requires an input of $0.40 from each sector.
Use this matrix to solve Problem 27. Discuss any differences Find the output for each sector that is needed to satisfy a final
in your calculations and in your answers. demand of $5 billion for transportation and $20 billion for
manufacturing.
30. Refer to Problem 28. Fill in the elements in the following
37. Agriculture, tourism. The economy of a small island nation
technology matrix.
is based on two sectors, agriculture and tourism. Production
T E of a dollar’s worth of agriculture requires an input of $0.20
T from agriculture and $0.20 from tourism. Production of a
T = c d
E dollar’s worth of tourism requires an input of $0.50 from ag-
Use this matrix to solve Problem 28. Discuss any differences riculture and $0.25 from tourism. Find the output from each
in your calculations and in your answers. sector that is needed to satisfy a final demand of $50 million
for agriculture and $90 million for tourism.
31. The technology matrix for an economy based on energy (E)
and mining (M) is 38. Agriculture, oil. The economy of a country is based on two
sectors, agriculture and oil. Production of a dollar’s worth of
E M
agriculture requires an input of $0.40 from agriculture and
E 0.2 0.3
M = c d $0.35 from oil. Production of a dollar’s worth of oil requires
M 0.4 0.3
an input of $0.20 from agriculture and $0.05 from oil. Find (in billions of dollars) for a 3-year period. Find the output for
the output from each sector that is needed to satisfy a final de- each sector that is needed to satisfy each of these final demands.
mand of $40 million for agriculture and $250 million for oil. Round answers to the nearest billion dollars.
39. Agriculture, manufacturing, energy. An economy is based Output Final Demand
on three sectors, agriculture, manufacturing, and energy. A E L M 1 2 3
Production of a dollar’s worth of agriculture requires inputs A 0.05 0.17 0.23 0.09 23 32 55
of $0.20 from agriculture, $0.20 from manufacturing, and E 0.07 0.12 0.15 0.19 41 48 62
$0.20 from energy. Production of a dollar’s worth of manu- Input
L 0.25 0.08 0.03 0.32 18 21 25
facturing requires inputs of $0.40 from agriculture, $0.10
from manufacturing, and $0.10 from energy. Production of M 0.11 0.19 0.28 0.16 31 33 35
a dollar’s worth of energy requires inputs of $0.30 from ag- 42. Repeat Problem 41 with the following table:
riculture, $0.10 from manufacturing, and $0.10 from energy.
Output Final Demand
Find the output for each sector that is needed to satisfy a final
A E L M 1 2 3
demand of $10 billion for agriculture, $15 billion for manu-
facturing, and $20 billion for energy. A 0.07 0.09 0.27 0.12 18 22 37
E 0.14 0.07 0.21 0.24 26 31 42
40. Electricity, natural gas, oil. A large energy company Input
L 0.17 0.06 0.02 0.21 12 19 28
produces electricity, natural gas, and oil. The production
of a dollar’s worth of electricity requires inputs of $0.30 M 0.15 0.13 0.31 0.19 41 45 49
from electricity, $0.10 from natural gas, and $0.20 from oil.
Production of a dollar’s worth of natural gas requires inputs
of $0.30 from electricity, $0.10 from natural gas, and $0.20
from oil. Production of a dollar’s worth of oil requires inputs
of $0.10 from each sector. Find the output for each sector that 0.2 0 0.4 1.7 0.2 0.9
is needed to satisfy a final demand of $25 billion for electric- 1. (A) £ 0 0.1 0.2 § (B) £ 0.2 1.2 0.4 §
ity, $15 billion for natural gas, and $20 billion for oil. 0.4 0.2 0.2 0.9 0.4 1.8
41. Four sectors. An economy is based on four sectors, agri-
culture (A), energy (E), labor (L), and manufacturing (M). (C) $71 billion for coal, $26 billion for oil, and $67 billion
The table gives the input requirements for a dollar’s worth of for transportation
output for each sector, along with the projected final demand

3.1 Review: Systems of Linear Equations in Two Variables EXAMPLES
• The solution of a system is an ordered pair of real numbers that satisfies each equation in the system. Ex. 1, p. 153
Solution by graphing is one method that can be used to find a solution. Ex. 2, p. 154
• A linear system is consistent and independent if it has a unique solution, consistent and dependent if it
has more than one solution, and inconsistent if it has no solution. A linear system that is consistent and
dependent actually has an infinite number of solutions.
• A graphing calculator provides accurate solutions to a linear system. Ex. 3, p. 156
• The substitution method can also be used to solve linear systems. Ex. 4, p. 157
• The method of elimination by addition is easily extended to larger systems. Ex. 5, p. 158
3.2 Systems of Linear Equations and Augmented Matrices
• A matrix is a rectangular array of real numbers. Row operations performed on an augmented matrix Ex. 1, p. 169
produce equivalent systems (Theorem 1, page 168). Ex. 2, p. 171
• There are only three possible final forms for the augmented matrix for a linear system of two equations Ex. 3, p. 171
in two variables (p. 173). Ex. 4, p. 173

• There are many possibilities for the final reduced form of the augmented matrix of a larger system of Ex. 1, p. 176
linear equations. Reduced form is defined on page 176. Ex. 2, p. 177
• The Gauss–Jordan elimination procedure is described on pages 178 and 179. Ex. 3, p. 179
Ex. 4, p. 180
Ex. 5, p. 181
3.4 Matrices: Basic Operations
• Two matrices are equal if they are the same size and their corresponding elements are equal. The sum of Ex. 1, p. 189
two matrices of the same size is the matrix with elements that are the sum of the corresponding elements
of the two given matrices.
• The negative of a matrix is the matrix with elements that are the negatives of the given matrix. If A and Ex. 2, p. 190
B are matrices of the same size, then B can be subtracted from A by adding the negative of B to A.
• Matrix equations involving addition and subtraction are solved much like real number equations. Ex. 3, p. 190
• The product of a real number k and a matrix M is the matrix formed by multiplying each element of M by k. Ex. 4, p. 191
• The product of a row matrix and a column matrix is defined on page 192. Ex. 6, p. 192
• The matrix product of an m * p matrix with a p * n is defined on page 193. Ex. 8, p. 194
3.5 Inverse of a Square Matrix
• The identity matrix for multiplication is defined on page 202. Ex. 1, p. 202
• The inverse of a square matrix is defined on page 204. Ex. 2, p. 205
Ex. 3, p. 208
Ex. 4, p. 209
3.6 Matrix Equations and Systems of Linear Equations
• Basic properties of matrices are summarized in Theorem 1 on page 214. Ex. 1, p. 215
• Matrix inverse methods for solving systems of equations are described in the Summary on page 217. Ex. 2, p. 215
Ex. 3, p. 216
3.7 Leontief Input–Output Analysis
• Leontief input–output analysis is summarized on page 225. Ex. 1, p. 226
Review Exercises
Work through all the problems in this chapter review and check 1 0 1 2 1 1 0 2
your answers in the back of the book. Answers to all problems are (C) c 2 d (D) c 2 d
0 1 1 3 0 1 1 3
there along with section numbers in italics to indicate where each
type of problem is discussed. Where weaknesses show up, review 4. Given matrices A and B,
appropriate sections in the text. -3 2
1. Solve the following system by graphing: 5 3 -1 0 2
A = c d B = C 0 4S
2x - y = 4 -4 8 1 3 0
-1 7
x - 2y = - 4 (A) What is the size of A? Of B?
2. Solve the system in Problem 1 by substitution. (B) Find a24, a15, b31, and b22.
3. If a matrix is in reduced form, say so. If not, explain why and (C) Is AB defined? Is BA defined?
state the row operation(s) necessary to transform the matrix 5. Find x1 and x2:
into reduced form.
1 -2 x 4
(A) c dc 1 d = c d
0 1 2 1 0 2 1 -3 x2 2
(A) c 2 d (B) c 2 d
1 0 3 0 3 3 5 3 x 25 18
(B) c dc 1 d + c d = c d
1 1 x2 14 22
In Problems 6–14, perform the operations that are defined, given 27. Solve the system in Problem 26A by writing the system as a
the following matrices: matrix equation and using the inverse of the coefficient matrix
(see Problem 25). Also, solve the system if the constants 1, 3,
1 2 2 1 1 and 3 are replaced by 0, 0, and - 2, respectively. By - 3, - 4,
A = c d B = c d C = [2 3] D = c d
3 1 1 1 2 and 1, respectively.
6. A + 2B 7. 3B + D 28. Discuss the relationship between the number of solutions of
the following system and the constant k.
8. 2A + B 9. BD
2x1 - 6x2 = 4
10. BC 11. AD
- x1 + kx2 = - 2
12. DC 13. CA
29. An economy is based on two sectors, agriculture and energy.
14. C + A Given the technology matrix M and the final demand matrix D
15. Find the inverse of the matrix A given below by appropriate (in billions of dollars), find 1I - M2 -1 and the output matrix X:
row operations on [A ∙ I ]. Show that A - 1A = I. A E
A 0.2 0.15 A 30
4 3 M = c d D = c d
A = c d E 0.4 0.3 E 20
3 2
30. Use the matrix M in Problem 29 to fill in the elements in the
16. Solve the following system using elimination by addition:
following technology matrix.
3m1 + 4m2 = 3 E A
2m1 + 3m2 = 3 E
T = c d
A
17. Solve the system in Problem 16 by performing appropriate
row operations on the augmented matrix of the system. Use this matrix to solve Problem 29. Discuss any differences
in your calculations and in your answers.
18. Solve the system in Problem 16 by writing the system as a
matrix equation and using the inverse of the coefficient ma- 31. An economy is based on two sectors, coal and steel. Given
trix (see Problem 15). Also, solve the system if the constants the technology matrix M and the final demand matrix D (in
3 and 5 are replaced by 7 and 10, respectively. By 4 and 2, billions of dollars), find 1I - M2 -1 and the output matrix X:
respectively. C S
C 0.45 0.65 C 40
In Problems 19–24, perform the operations that are defined, given M = c d D = c d
S 0.55 0.35 S 10
the following matrices:
32. Use graphical approximation techniques on a graphing
2 -2 -1 calculator to find the solution of the following system to two
A = C1 0S B = C 2S C = [2 1 3] decimal places:
3 2 3
x - 5y = - 5
3 -2 1 3 -4 2x + 3y = 12
D = c d E = c d
-1 1 2 -1 0
33. Find the inverse of the matrix A given below. Show that
19. A + B 20. E + DA A-1A = I.
21. DA - 3E 22. BC 4 5 6
A = C 4 5 -4 S
23. CB 24. AD - BC
1 1 1
25. Find the inverse of the matrix A given below by appropriate
row operations on [A ∙ I]. Show that A - 1A = I. 34. Solve the system
1 2 3 0.04x1 + 0.05x2 + 0.06x3 = 360

A = C2 3 4S 0.04x1 + 0.05x2 - 0.04x3 = 120
1 2 1 x1 + x2 + x3 = 7,000
26. Solve by Gauss–Jordan elimination:
by writing it as a matrix equation and using the inverse of the
(A) x1 + 2x2 + 3x3 = 1 coefficient matrix. (Before starting, multiply the first two equa-
2x1 + 3x2 + 4x3 = 3 tions by 100 to eliminate decimals. Also, see Problem 33.)
x1 + 2x2 + x3 = 3 35. Solve Problem 34 by Gauss–Jordan elimination.
(B) x1 + 2x2 - x3 = 2 36. Given the technology matrix M and the final demand matrix D
2x1 + 3x2 + x3 = - 3 (in billions of dollars), find 1I - M2 -1 and the output matrix X:
3x1 + 5x2 = -1 0.2 0 0.4 40
(C) x1 + x2 + x3 = 8 M = C 0.1 0.3 0.1 S D = C 20 S
3x1 + 2x2 + 4x3 = 21 0 0.4 0.2 30
37. Discuss the number of solutions for a system of n equations (B) Solve Problem 41 as in part (A) if 7.5 tons of nickel and
in n variables if the coefficient matrix 7 tons of copper are needed.
(A) Has an inverse. 43. Business leases. A grain company wants to lease a fleet
(B) Does not have an inverse. of 20 covered hopper railcars with a combined capacity of
38. Discuss the number of solutions for the system corresponding 108,000 cubic feet. Hoppers with three different carrying
to the reduced form shown below if capacities are available: 3,000 cubic feet, 4,500 cubic feet,
and 6,000 cubic feet.
(A) m ∙ 0
(A) How many of each type of hopper should they lease?
(B) m = 0 and n ∙ 0
(C) m = 0 and n = 0 (B) The monthly rates for leasing these hoppers are $180
for 3,000 cubic feet, $225 for 4,500 cubic feet, and $325
1 0 -2 5 for 6,000 cubic feet. Which of the solutions in part (A)
C0 1 3 3 3S would minimize the monthly leasing costs?
0 0 m n 44. Material costs. A manufacturer wishes to make two differ-
39. One solution to the input–output equation X = MX + D is ent bronze alloys in a metal foundry. The quantities of cop-
given by X = 1I - M2 -1D. Discuss the validity of each step per, tin, and zinc needed are indicated in matrix M. The costs
in the following solutions of this equation. (Assume that all for these materials (in dollars per pound) from two suppliers
necessary inverses exist.) Are both solutions correct? are summarized in matrix N. The company must choose one
(A) X = MX + D supplier or the other.
X - MX = D Copper Tin Zinc
X1I - M2 = D 4,800 lb 600 lb 300 lb Alloy 1
M = c d
X = D1I - M2 -1 6,000 lb 1,400 lb 700 lb Alloy 2
(B) X = MX + D Supplier A Supplier B
- D = MX - X $0.75 $0.70 Copper
- D = 1M - I2X N = C $6.50 $6.70 S Tin
X = 1M - I2 -1 1 - D2 $0.40 $0.50 Zinc
(A) Discuss possible interpretations of the elements in the
matrix products MN and NM.
Applications (B) If either product MN or NM has a meaningful interpreta-

tion, find the product and label its rows and columns.
40. Break-even analysis. A computer design company is pre- (C) Discuss methods of matrix multiplication that can be
paring to market a new computer. The company’s fixed costs used to determine the supplier that will provide the nec-
for research, development, and tooling are $253,000 and the essary materials at the lowest cost.
variable costs are $35.50 per unit. The company is planning
to sell each unit at a market price of $199.99. 45. Labor costs. A company with manufacturing plants in
(A) Find the cost and revenue equations. Tokyo and Osaka has labor-hour and wage requirements for
the manufacture of two inexpensive calculators as given in
(B) Find the break-even point. matrices M and N below:
(C) Graph both equations in the same coordinate system and Labor-hours per calculator
show the break-even point. Use the graph to determine Fabricating Assembly Packaging
the production levels that will result in a profit and in a department department department
loss. 0.15 hr 0.10 hr 0.05 hr Model A
M = c d
41. Resource allocation. An international mining company has 0.25 hr 0.20 hr 0.05 hr Model B
two mines in Port Hedland and Jack Hills. The composition Hourly wages
of the ore from each field is given in the table. How many Tokyo Osaka
tons of ore from each mine should be used to obtain exactly 6 plant plant
tons of nickel and 8 tons of copper?
$12 $10 Fabricating department
Mine Nickel (%) Copper (%) N = C $15 $12 S Assembly department
Port Hedland 2 4 $ 7 $ 6 Packaging department
Jack Hills 3 2
(A) Find the labor cost for producing one model B calculator
42. Resource allocation. at the Tokyo plant.
(A) Set up Problem 41 as a matrix equation and solve using (B) Discuss possible interpretations of the elements in the
the inverse of the coefficient matrix. matrix products MN and NM.
(C) If either product MN or NM has a meaningful interpreta- (B) Find the output of each item that is needed to satisfy
tion, find the product and label its rows and columns. a final demand of 800 metric tons of wheat and 2000
metric tons of oil.
46. Investment analysis. A person has $5,000 to invest, part
at 5% and the rest at 10%. How much should be invested 52. Cryptography. The following message was encoded with
at each rate to yield $400 per year? Solve using augmented the matrix B shown below. Decode the message.
matrix methods.
7 25 30 19 6 24 20 8 28 5 14 14
47. Investment analysis. Solve Problem 46 by using a matrix 9 23 28 15 6 21 13 1 14 21 26 29
equation and the inverse of the coefficient matrix.
1 1 0
48. Investment analysis. In Problem 46, is it possible to have an B = C1 0 1S
annual yield of $200? Of $600? Describe all possible annual 1 1 1
yields.
49. Ticket prices. An outdoor amphitheater has 25,000 seats. 53. Traffic flow. The rush-hour traffic flow (in vehicles per
Ticket prices are $8, $12, and $20, and the number of tickets hour) for a network of four one-way streets is shown in the
priced at $8 must equal the number priced at $20. How many figure.
tickets of each type should be sold (assuming that all seats
1st St. 2nd St.
can be sold) to bring in each of the returns indicated in the
table? Solve using the inverse of the coefficient matrix. 700 200
Concert Elm St.
1 2 3 600 x1 600
x4 x2
Tickets sold 25,000 25,000 25,000
Return required $320,000 $330,000 $340,000 Oak St.
300 x3 300
500 400
50. Ticket prices. Discuss the effect on the solutions to Problem 49
if it is no longer required to have an equal number of $8 tickets
and $20 tickets.
(A) Write the system of equations determined by the flow of
51. Input–output analysis. Consider a hypothetical economy traffic through the four intersections.
that depends on two basic products, wheat and oil. Production
(B) Find the solution of the system in part (A).
of 1 metric ton of wheat requires an input of 0.25 metric tons
of wheat and 0.33 metric tons of oil. Production of 1 metric (C) What is the maximum number of vehicles per hour that
ton of oil requires 0.08 metric tons of wheat and 0.11 metric can travel from Oak Street to Elm Street on 1st Street?
tons of oil. What is the minimum number?
(A) Find the output of each item that is needed to satisfy (D) If traffic lights are adjusted so that 500 vehicles per hour
a final demand of 500 metric tons of wheat and 1000 travel from Oak Street to Elm Street on 1st Street, deter-
metric tons of oil. mine the flow around the rest of the network.
4 Linear Inequalities and
Linear Programming
4.1 Linear Inequalities in Two
Variables
Introduction
Real-world problems often involve limitations on materials, time, and money.
4.2 Systems of Linear
Inequalities in Two To express such constraints mathematically, we formulate systems of inequali-
Variables ties. In Chapter 4 we discuss systems of inequalities in two variables and
introduce a relatively new mathematical tool called linear programming. Linear
4.3 Linear Programming
programming can be used to determine how two different alloys of bronze
in Two Dimensions:
A Geometric Approach should be combined to produce window frames (see Problems 59 and 60 in
Section 4.1).
235
236 CHAPTER 4 Linear Inequalities and Linear Programming
4.1 Linear Inequalities in Two Variables

■■ Graphing Linear Inequalities Graphing Linear Inequalities in Two Variables
in Two Variables
We know how to graph first-degree equations such as
■■ Application
y = 2x - 3 and 2x - 3y = 5
but how do we graph first-degree inequalities such as the following?
y … 2x - 3 and 2x - 3y 7 5
We will find that graphing these inequalities is similar to graphing the equations,
but first we must discuss some important subsets of a plane in a rectangular coordi-
nate system.
A line divides the plane into two regions called half-planes. A vertical line
divides it into left and right half-planes; a nonvertical line divides it into upper and
lower half-planes. In either case, the dividing line is called the boundary line of
each half-plane, as indicated in Figure 1.
Boundary Boundary
y y
line line
Left Right Upper

half-plane half-plane half-plane
x x
Lower
half-plane
(A) (B)
Figure 1
To find the half-planes determined by a linear equation such as y - x = - 2,

we rewrite the equation as y = x - 2. For any given value of x, there is exactly
one value for y such that (x, y) lies on the line. For example, for x = 4, we have
y = 4 - 2 = 2. For the same x and smaller values of y, the point (x, y) will lie below
the line since y 6 x - 2. So the lower half-plane corresponds to the solution of the
inequality y 6 x - 2. Similarly, the upper half-plane corresponds to y 7 x - 2,
as shown in Figure 2.
y
y5x22
(4, y)
y . 4 2 2:
point in upper half-plane
(4, y)
y 5 4 2 2: point on boundary line
x
4
(4, y)
y , 4 2 2: point in lower half-plane
Figure 2
SECTION 4.1 Linear Inequalities in Two Variables 237
The four inequalities formed from y = x - 2, replacing the = sign by 7, Ú ,6 ,

and … , respectively, are
y 7 x - 2 y Ú x - 2 y 6 x - 2 y … x - 2
The graph of each is a half-plane, excluding the boundary line for 6 and 7 and
including the boundary line for … and Ú . In Figure 3, the half-planes are indicated
with small arrows on the graph of y = x - 2 and then graphed as shaded regions.
Excluded boundary lines are shown as dashed lines, and included boundary lines are
shown as solid lines.
y y y y
5 5 5 5
y,x22 y#x22
x x x x
25 5 25 5 25 5 25 5
y.x22 y$x22
25 25 25 25
(A) (B) (C) (D)

Figure 3
Figure 4 shows the graphs of Figures 3B and 3D on a graphing calculator.
Note that it is impossible to show a dotted boundary line when using shading on a
calculator.
5 5
25 5 25 5
25 25
(A) (B)
Figure 4
The preceding discussion suggests the following theorem, which is stated with-
out proof:
THEOREM 1 Graphs of Linear Inequalities

The graph of the linear inequality
Ax + By 6 C or Ax + By 7 C
with B ∙ 0, is either the upper half-plane or the lower half-plane (but not both)
determined by the line Ax + By = C.
If B = 0 and A ∙ 0, the graph of
Ax 6 C or Ax 7 C
is either the left half-plane or the right half-plane (but not both) determined by the
line Ax = C.
As a consequence of this theorem, we state a simple and fast mechanical proce-

dure for graphing linear inequalities.
PROCEDURE Graphing Linear Inequalities

Step 1 First graph Ax + By = C as a dashed line if equality is not included in
the original statement, or as a solid line if equality is included.
Step 2 Choose a test point anywhere in the plane not on the line [the origin 10, 02
usually requires the least computation], and substitute the coordinates into
the inequality.
Step 3 Does the test point satisfy the original inequality? If so, shade the half-
plane that contains the test point. If not, shade the opposite half-plane.
EXAMPLE 1 Graphing a Linear Inequality Graph 2x - 3y … 6.

SOLUTION
Reminder Step 1 Graph 2x - 3y = 6 as a solid line, since equality is included in the original
Recall that the line 2x - 3y = 6 can statement (Fig. 5).
be graphed by finding any two points
on the line. The x and y intercepts are y y
usually a good choice (see Fig. 5). 10 10
x y
0 -2 2x 2 3y 5 6
5 5
3 0
x x
210 25 5 10 210 25 5 10
25 25
210 210
Figure 5 Figure 6
Step 2 Pick a convenient test point above or below the line. The origin 10, 02 requires
the least computation, so substituting 10, 02 into the inequality, we get
2x - 3y … 6
2102 - 3102 = 0 … 6
This is a true statement; therefore, the point 10, 02 is in the solution set.
Step 3 The line 2x - 3y = 6 and the half-plane containing the origin form the
graph of 2x - 3y … 6, as shown in Figure 6.
Matched Problem 1 Graph 6x - 3y 7 18.

In Step 2 of Example 1, 10, 02 was used as a test point in graphing a linear inequal-
ity. Describe those linear inequalities for which 10, 02 is not a valid test point. In that
case, how would you choose a test point to make calculation easy?
EXAMPLE 2 Graphing Inequalities Graph

(A) y 7 - 3 (B) 2x … 5 (C) x … 3y
SOLUTION
(A) Step 1 Graph the horizontal line y = - 3 as a dashed line, since equality is not
included in the original statement (Fig. 7).
Step 2 Substituting x = 0 and y = 0 in the inequality produces a true state-

ment, so the point 10, 02 is in the solution set.
Step 3 The graph of the solution set is the upper half-plane, excluding the
boundary line (Fig. 8).
y y
5 5
x x
25 0 5 25 0 5
25 25
Figure 7 Figure 8
(B) Step 1 Graph the vertical line 2x = 5 as a solid line, since equality is included
in the original statement (Fig. 9).
Step 2 Substituting x = 0 and y = 0 in the inequality produces a true state-
ment, so the point 10, 02 is in the solution set.
Step 3 The graph of the solution set is the left half-plane, including the bound-
ary line (Fig. 10).
y y
5 5
x x
25 0 5 25 0 5
25 25
Figure 9 Figure 10
(C) Step 1 Graph the line x = 3y as a solid line, since equality is included in the
original statement (Fig. 11).
Step 2 Since the line passes through the origin, we must use a different test
point. We choose 10, 22 for a test point and conclude that this point is
in the solution set.
Step 3 The graph of the solution set is the upper half-plane, including the
boundary line (Fig. 12).
y y
5 5
x x
25 0 5 25 0 5
25 25
Figure 11 Figure 12
Matched Problem 2 Graph

(A) y 6 4 (B) 4x Ú - 9 (C) 3x Ú 2y
EXAMPLE 3 Interpreting a Graph Find the linear inequality whose graph is given in Figure 13.
Write the boundary line equation in the form Ax + By = C, where A, B, and C are
integers, before stating the inequality.
y SOLUTION The boundary line (Fig. 13) passes through the points 10, 62 and 18, 02.
10 We use the slope-intercept form to find the equation of this line:
0 - 6 6 3
Slope: m = = - = -
8 - 0 8 4
5 y intercept: b = 6
3
Boundary line equation: y = - x + 6 Multiply both sides by 4.
4
x
5 10 4y = - 3x + 24 Add 3x to both sides.
3x + 4y = 24 Form: Ax + By = C
Figure 13 Since 10, 02 is in the shaded region in Figure 13 and the boundary line is solid, the
graph in Figure 13 is the graph of 3x + 4y … 24.
Matched Problem 3 Find the linear inequality whose graph is given in Figure 14.
Write the boundary line equation in the form Ax + By = C, where A, B, and C are
integers, before stating the inequality.
y
10
x
5 10
Figure 14
Application
EXAMPLE 4 Sales A concert promoter wants to book a rock group for a stadium concert. A
ticket for admission to the stadium playing field will cost $125, and a ticket for
a seat in the stands will cost $175. The group wants to be guaranteed total ticket
sales of at least $700,000. How many tickets of each type must be sold to satisfy
the group’s guarantee? Express the answer as a linear inequality and draw its graph.
SOLUTION
Let x = Number of tickets sold for the playing field
y = Number of tickets sold for seats in the stands
We use these variables to translate the following statement from the problem into a
mathematical statement:
The group wants to be guaranteed total ticket sales of at least $700,000.
Sales for the Sales for seats At Total sales
a b + a b a b a b
playing field in the stands least guaranteed
125x + 175y Ú 700,000
Dividing both sides of this inequality by 25, x, and y must satisfy
5x + 7y Ú 28,000
y We use the three-step procedure to graph this inequality.

10,000 Step 1 Graph 5x + 7y = 28,000 as a solid line (Fig. 15).
Step 2 Substituting x = 0 and y = 0 in the inequality produces a false statement,
so the point 10, 02 is not in the solution set.
5,000 Step 3 The graph of the inequality is the upper half-plane including the boundary
line (Fig. 16), but does this graph really represent ticket sales?
y y
x
5,000 10,000
10,000 10,000
Figure 15
5,000 5,000
x x
5,000 10,000 5,000 10,000
Figure 16 Figure 17
The shaded region in Figure 16 contains points in the second quadrant (where
x 6 0) and the fourth quadrant (where y 6 0). It is not possible to sell a negative
number of tickets, so we must restrict both x and y to the first quadrant. With this
restriction, the solution becomes
5x + 7y Ú 28,000
x Ú 0, y Ú 0
and the graph is shown in Figure 17. There is yet another restriction on x and y. It is
not possible to sell a fractional number of tickets, so both x and y must be integers.
So the solutions of the original problem correspond to those points of the shaded
region in Figure 17 that have integer coordinates. This restriction is not indicated in
Figure 17, because the points with integer coordinates are too close together (about
9,000 such points per inch) to be visually distinguishable from other points.
Matched Problem 4 A food vendor at a rock concert sells sandwiches for $4

and fish and chips for $5. How many of these must be sold to produce sales of at
least $1,000? Express the answer as a linear inequality and draw its graph.
Exercises 4.1
Skills Warm-up Exercises A Graph each inequality in Problems 9–18.
W For Problems 1–8, if necessary, review Section A.2. 9. y … x - 1 10. y 7 x + 1

1. Is the point 13, 52 on the line y = 2x + 1? 11. 3x - 2y 7 6 12. 2x - 5y … 10
13. x Ú - 4 14. y 6 5
2. Is the point 17, 92 on the line y = 3x - 11?
15. 5x + 7y Ú 35 16. 4x + 8y Ú 32
3. Is the point 13, 52 in the solution set of y … 2x + 1? 17. 5x … - 2y 18. 6x Ú 4y
4. Is the point 17, 92 in the solution set of y … 3x - 11?
In Problems 19–22,
5. Is the point 110, 122 on the line 13x - 11y = 2?
(A) graph the set of points that satisfy the inequality.
6. Is the point 121, 252 on the line 30x - 27y = 1?
(B) graph the set of points that do not satisfy the inequality.
7. Is the point 110, 122 in the solution set of 13x - 11y Ú 2?
19. 3x + 5y 6 30 20. 3x + 4y 7 24
8. Is the point 121, 252 in the solution set of 30x - 27y … 1?
21. 5x - 2y Ú 20 22. 3x - 5y … 30
In Problems 23–32, define the variable and translate the sentence 40. New-car sales and used-car sales combined are at most
into an inequality. $500,000.
23. There are fewer than 10 applicants. 41. Revenue is at least $20,000 under the cost.
24. She consumes no more than 900 calories per day. 42. The Conservative candidate beat the Labor candidate by at
least seven percentage points.
25. He practices no less than 2.5 hours per day.
43. The number of grams of protein in rice is more than double
26. The average attendance is less than 15,000.
the number of grams of protein in potato.
27. The success rate of the project is under 20%.
44. The plane is at least 500 miles closer to Cape Town than to
28. The discount is at least 5%. Johannesburg.
29. The annual fee is over $45,000. C In Problems 45–54, graph each inequality subject to the non-
30. The population is greater than 500,000. negative restrictions.
31. The enrollment is at most 30. 45. 15x + 50y … 4,500, x Ú 0, y Ú 0
32. Mileage exceeds 35 miles per gallon. 46. 24x + 30y 7 7,200, x Ú 0, y Ú 0
B In Exercises 33–38, state the linear inequality whose graph is 47. 60x - 35y 7 0, x Ú 0, y Ú 0
given in the figure. Write the boundary-line equation in the form 48. 16x - 12y Ú 4,800, x Ú 0, y Ú 0
Ax + By = C, where A, B, and C are integers, before stating the
inequality. 49. - 18x + 30y Ú 2,700, x Ú 0, y Ú 0
33. y 34. y 50. - 14x + 22y 6 1,540, x Ú 0, y Ú 0
5 5 51. 12x - 8y 6 4,800, x Ú 0, y Ú 0

52. - 35x + 75y … 0, x Ú 0, y Ú 0
53. 25x + 75y 6 - 600, x Ú 0, y Ú 0
x x
25 0 5 25 0 5
54. 75x + 25y 7 - 600, x Ú 0, y Ú 0
25 25
35. y 36. y Applications

5 5
In Problems 55–66, express your answer as a linear inequality
with appropriate nonnegative restrictions and draw its graph.
55. Seed costs. Seed costs for a farmer are $90 per acre for corn
x x
25 0 5 25 0 5 and $70 per acre for soybeans. How many acres of each crop
should the farmer plant if he wants to spend no more than
$11,000 on seed?
56. Labor costs. Labor costs for a farmer are $120 per acre for
25 25
corn and $100 per acre for soybeans. How many acres of
37. y 38. y
each crop should the farmer plant if he wants to spend no
5 5 more than $15,000 on labor?
57. Fertilizer. A farmer wants to use two brands of fertilizer for
his corn crop. Brand A contains 26% nitrogen, 3% phosphate,
0
x
0
x and 3% potash. Brand B contains 16% nitrogen, 8% phos-
25 5 25 5
phate, and 8% potash.
(Source: Spectrum Analytic, Inc.)
25 25 (A) How many pounds of each brand of fertilizer should he
add to each acre if he wants to add at least 120 pounds of
In Problems 39–44, define two variables and translate the sen- nitrogen to each acre?
tence into an inequality. (B) How many pounds of each brand of fertilizer should he
39. Enrollment in finite mathematics plus enrollment in calculus add to each acre if he wants to add at most 28 pounds of
is less than 300. phosphate to each acre?
58. Fertilizer. A farmer wants to use two brands of fertilizer for Answers to Matched Problems
his soybean crop. Brand A contains 18% nitrogen, 24% phos-
phate, and 12% potash. Brand B contains 5% nitrogen, 10% 1. y
phosphate, and 15% potash.
5
(Source: Spectrum Analytic, Inc.)
(A) How many pounds of each brand of fertilizer should he
add to each acre if he wants to add at least 50 pounds of
x
phosphate to each acre? 25 0 5
(B) How many pounds of each brand of fertilizer should he

add to each acre if he wants to add at most 60 pounds of
25
potash to each acre?
59. Alloy. A producer uses two alloys of bronze to produce
window frames: commercial bronze that is 90% copper
and 10% zinc and architectural bronze that is 57% copper,
40% zinc, and 3% lead. How many kilograms of each alloy 2. (A)
y
should the producer use to produce a frame that is at least
75% copper? 5
60. Alloy. Refer to Exercise 59. How many kilograms of each

alloy should the producer use to produce a frame that is at
least 30% zinc? 0
x
25 5
61. Customized vehicles. A company uses sedans and minivans
to produce custom vehicles for transporting hotel guests
to and from airports. Plant A can produce 10 sedans and 25
8 minivans per week, and Plant B can produce 8 sedans and
6 minivans per week. How many weeks should each plant (A)
operate in order to produce at least 400 sedans? (B) (C)
y y
62. Customized vehicles. Refer to Exercise 61. How many
5 5
weeks should each plant operate in order to produce at least
480 minivans?
63. Political advertising. A candidate has budgeted $10,000 to
x x
spend on radio and television advertising. A radio ad costs 25 0 5 25 0 5
$200 per 30-second spot, and a television ad costs $800 per
30-second spot. How many radio and television spots can the
candidate purchase without exceeding the budget? 25 25
64. Political advertising. Refer to Problem 63. The candidate

(B) (C)
decides to replace the television ads with newspaper ads that
cost $500 per ad. How many radio spots and newspaper ads 3. 2x + 5y 7 20
can the candidate purchase without exceeding the budget? 4. Let x = Number of hot dogs sold
y = Number of hamburgers sold
65. Mattresses. A company produces foam mattresses in two
4x + 5y Ú 1,000 x Ú 0, y Ú 0
sizes: regular and king. It takes 5 minutes to cut the foam for
a regular mattress and 6 minutes for a king mattress. If the y
cutting department has 50 labor-hours available each day, 400
how many regular and king mattresses can be cut in one day?
300
66. Mattresses. Refer to Problem 65. It takes 15 minutes to
200
cover a regular mattress and 20 minutes to cover a king mat-
tress. If the covering department has 160 labor-hours avail- 100
able each day, how many regular and king mattresses can be x
covered in one day? 100 200 300 400
4.2 Systems of Linear Inequalities in Two Variables

■■ Solving Systems of Linear Solving Systems of Linear Inequalities Graphically
Inequalities Graphically
We now consider systems of linear inequalities such as
■■ Applications
x + y Ú 6 and 2x + y … 22
2x - y Ú 0 x + y … 13
2x + 5y … 50
x Ú 0
y Ú 0
We wish to solve such systems graphically—that is, to find the graph of all ordered
pairs of real numbers (x, y) that simultaneously satisfy all the inequalities in the
system. The graph is called the solution region for the system (the solution region
is also known as the feasible region). To find the solution region, we graph each
inequality in the system and then take the intersection of all the graphs. To simplify
the discussion that follows, we consider only systems of linear inequalities where
equality is included in each statement in the system.
EXAMPLE 1 Solving a System of Linear Inequalities Graphically Solve the following system
y of linear inequalities graphically:
x + y Ú 6
5 2x - y Ú 0
SOLUTION Graph the line x + y = 6 and shade the region that satisfies the
linear inequality x + y Ú 6. This region is shaded with red lines in Fig-
x
5 ure 1A. Next, graph the line 2x - y = 0 and shade the region that satisfies
the inequality 2x - y Ú 0. This region is shaded with blue lines in Figure
x1y56 1A. The solution region for the system of inequalities is the intersection of
2x 2 y 5 0
(A) these two regions. This is the region shaded in both red and blue (cross-
y hatched) in Figure 1A and redrawn in Figure 1B with only the solution re-
gion shaded. The coordinates of any point in the shaded region of Figure
Solution 1B specify a solution to the system. For example, the points 12, 42, 16, 32,
5 region and 17.43, 8.562 are three of infinitely many solutions, as can be easily
(2, 4) checked. The intersection point 12, 42 is obtained by solving the equations
x + y = 6 and 2x - y = 0 simultaneously using any of the techniques dis-
cussed in Chapter 3.
x
5
Matched Problem 1 Solve the following system of linear inequalities
graphically:
x1y56 3x + y … 21
2x 2 y 5 0
(B)
x - 2y … 0
Figure 1
CONCEPTUAL INSIGHT
To check that you have shaded a solution region correctly, choose a test point in
the region and check that it satisfies each inequality in the system. For example,
choosing the point 15, 42 in the shaded region in Figure 1B, we have
x + y Ú 6 2x - y Ú 0
? ?
5 + 4 Ú 6 10 - 4 Ú 0
✓ ✓
9 Ú 6 6 Ú 0
SECTION 4.2 Systems of Linear Inequalities in Two Variables 245
The points of intersection of the lines that form the boundary of a solution region
will play a fundamental role in the solution of linear programming problems, which
are discussed in the next section.
DEFINITION Corner Point

A corner point of a solution region is a point in the solution region that is the
intersection of two boundary lines.
For example, the point 12, 42 is the only corner point of the solution region in
Example 1 (Fig. 1B).
EXAMPLE 2 Solving a System of Linear Inequalities Graphically Solve the following system
of linear inequalities graphically and find the corner points:
2x + y … 22
x + y … 13
2x + 5y … 50
x Ú 0
y Ú 0
SOLUTION The inequalities x Ú 0 and y Ú 0 indicate that the solution region will
lie in the first quadrant. So we can restrict our attention to that portion of the plane.
First, we graph the lines
2x + y = 22 Find the x and y intercepts of each line;
then sketch the line through these points.
x + y = 13
2x + 5y = 50
Next, choosing 10, 02 as a test point, we see that the graph of each of the first three
inequalities in the system consists of its corresponding line and the half-plane lying
below the line, as indicated by the small arrows in Figure 2. The solution region of
the system consists of the points in the first quadrant that simultaneously lie on or
below all three of these lines (see the shaded region in Fig. 2).
y
2x 1 y 5 22
25
20
x 1 y 5 13
(0, 10)
15
2x 1 5y 5 50 (5, 8)
10 (9, 4)
(11, 0)
5
x
(0, 0) 5 10 15 20 25
Figure 2
The corner points 10, 02, 10, 102, and 111, 02 can be determined from the
graph. The other two corner points are determined as follows:
Solve the system Solve the system

2x + 5y = 50 2x + y = 22
x + y = 13 x + y = 13
to obtain 15, 82. to obtain 19, 42.
Note that the lines 2x + 5y = 50 and 2x + y = 22 also intersect, but the

intersection point is not part of the solution region and so is not a corner point.
Matched Problem 2 Solve the following system of linear inequalities

graphically and find the corner points:
5x + y Ú 20
x + y Ú 12
x + 3y Ú 18
x Ú 0
y Ú 0
If we compare the solution regions of Examples 1 and 2, we see that there is a

fundamental difference between these two regions. We can draw a circle around the
solution region in Example 2; however, it is impossible to include all the points in the
solution region in Example 1 in any circle, no matter how large we draw it. This leads
to the following definition:
DEFINITION Bounded and Unbounded Solution Regions

A solution region of a system of linear inequalities is bounded if it can be enclosed
within a circle. If it cannot be enclosed within a circle, it is unbounded.
The solution region for Example 2 is bounded, and the solution region for
Example 1 is unbounded. This definition will be important in the next section.

Determine whether the solution region of each system of linear inequalities is
bounded or unbounded.
(A) y … 1 (B) x … 100 (C) x … y
x Ú 0 y … 200 y … x
y Ú 0 x Ú 0 x Ú 0
y Ú 0 y Ú 0
Applications
EXAMPLE 3 Nutrition A patient on a brown rice and skim milk diet is required to have at least
800 calories and at least 32 grams of protein per day. Each serving of brown rice
contains 200 calories and 5 grams of protein. Each serving of skim milk contains
80 calories and 8 grams of protein. How many servings of each food should be eaten
per day to meet the minimum daily requirements?
SOLUTION To answer the question, we need to solve for x and y, where
x = number of daily servings of brown rice

y = number of daily servings of skim milk
We arrange the information given in the problem in a table, with columns corre-
sponding to x and y.
Brown Rice Skim Milk Minimum Daily Requirement

Calories 200 cal/svg 80 cal/svg 800 cal
Protein 5 g/svg 8 g/svg 32 g
The number of calories in x servings of brown rice is 200x, and the number of calo-
ries in y servings of skim milk is 80y. So, to meet the minimum daily requirement
for calories, 200x + 80y must be greater than or equal to 800. This gives the first of
the inequalities below. The second inequality expresses the condition that the mini-
mum daily requirement for protein is met. The last two inequalities express the fact
that the number of servings of each food cannot be a negative number.
y
15
200x + 80y Ú 800 Requirement for calories
Solution 5x + 8y Ú 32 Requirement for protein
region
10 (0, 10) x Ú 0 Nonnegative restriction on x
y Ú 0 Nonnegative restriction on y
5
We graph this system of inequalities, and shade the solution region (Figure 3). Each
(3.2, 2)
(6.4, 0) point in the shaded area, including the straight-line boundaries, will meet the mini-
x mum daily requirements for calories and protein; any point outside the shaded area
25 0 5 10
will not. For example, 4 servings of brown rice and 2 servings of skim milk will
25 5x 1 8y 5 32 meet the minimum daily requirements, but 3 servings of brown rice and 2 servings
200x 1 80y 5 800 of skim milk will not. Note that the solution region is unbounded.
Figure 3
Matched Problem 3 A manufacturing plant makes two types of inflatable
boats—a two-person boat and a four-person boat. Each two-person boat requires
0.9 labor-hour in the cutting department and 0.8 labor-hour in the assembly depart-
ment. Each four-person boat requires 1.8 labor-hours in the cutting department and
1.2 labor-hours in the assembly department. The maximum labor-hours available
each month in the cutting and assembly departments are 864 and 672, respectively.
(A) Summarize this information in a table.
(B) If x two-person boats and y four-person boats are manufactured each month,
write a system of linear inequalities that reflects the conditions indicated.
Graph the feasible region.
Exercises 4.2
Skills Warm-up Exercises 7. Is the point 16, 22 in the solution region?
W For Problems 1–8, if necessary, review Section A.2. Problems 1–4 8. Is the point 15, 22 in the solution region?
refer to the following system of linear inequalities:
A In Problems 9–12, match the solution region of each system of
4x + y … 20 linear inequalities with one of the four regions shown in the figure.
3x + 5y … 37 y
3x 2 2y 5 0
x Ú 0 10
y Ú 0 II
1. Is the point 13, 52 in the solution region? 5 I
III (2, 3)
2. Is the point 14, 52 in the solution region?
3. Is the point 13, 62 in the solution region? x
210 25 5
4. Is the point 12, 62 in the solution region? x 1 2y 5 8
25 IV
Problems 5–8 refer to the following system of linear inequalities:
5x + y … 32 210
7x + 4y Ú 45
x Ú 0 9. x + 2y … 8 10. x + 2y Ú 8
y Ú 0 3x - 2y Ú 0 3x - 2y … 0
5. Is the point 14, 32 in the solution region? 11. x + 2y Ú 8 12. x + 2y … 8
3x - 2y Ú 0 3x - 2y … 0
6. Is the point 15, 32 in the solution region?
In Problems 13–16, solve each system of linear inequalities Solve the systems in Problems 29–38 graphically and indicate
graphically. whether each solution region is bounded or unbounded. Find the
13. 3x + y Ú 6 14. 3x + 4y … 12 coordinates of each corner point.
x … 4 y Ú -3
29. 2x + 3y … 12 30. 3x + 4y … 24
15. x - 2y … 12 16. 2x + 5y … 20 x Ú 0
x Ú 0
2x + y Ú 4 x - 5y Ú - 5 y Ú 0 y Ú 0
B In Problems 17–20, match the solution region of each system of
linear inequalities with one of the four regions shown in the figure. 31. 3x + 4 y Ú 24 32. 6x + 3y … 24
Identify the corner points of each solution region. x + y Ú 7 3x + 6y … 30
y x Ú 0 x Ú 0
y Ú 0 y Ú 0
33. 4x + y … 8 34. 4x + 3y Ú 24
(0, 16)
x + 4y … 17 3x + 4y Ú 8
I
x Ú 0 x Ú 0
II y Ú 0 y Ú 0
(0, 6) (6, 4)
III IV (18, 0) 35. 3x + 8y Ú 24 36. 3x + y … 21
x
(0, 0) (8, 0) x + y Ú 5 x + y … 9
x 1 3y 5 18
2x 1 y 5 16 3x + 2y Ú 12 x + 3y … 21
x Ú 0 x Ú 0
17. x + 3y … 18 18. x + 3y … 18
y Ú 0 y Ú 0
2x + y Ú 16 2x + y … 16
x Ú 0 x Ú 0 37. 5x + 4y Ú 48 38. 3x + y Ú 24
y Ú 0 y Ú 0 x + 2y Ú 12 x + y Ú 16
x + y Ú 11 x + 3y Ú 30
19. x + 3y Ú 18 20. x + 3y Ú 18
x Ú 0 x Ú 0
2x + y Ú 16 2x + y … 16
y Ú 0 y Ú 0
x Ú 0 x Ú 0
y Ú 0 y Ú 0 C Solve the systems in Problems 39–48 graphically and indicate
whether each solution region is bounded or unbounded. Find the
In Problems 21–28, is the solution region bounded or unbounded? coordinates of each corner point.
21. 5x - 4y … 10 22. x + 2y Ú 4 39. 2x - y … 1 40. x + y … 11

x Ú 0 x Ú 0 x + 2y … 13 x + 5y Ú 15
yÚ 0 y Ú 0 y Ú 15 2x + y Ú 12
23. 5x - 2y Ú 10 24. 4x - 3y … 12 41. x + 2y … 18 42. 2x + 3y Ú 24

x Ú 0 x Ú 0 2x + y … 21 x + 3y … 15
y Ú 0 y Ú 0 x + y Ú 10 y Ú 4
43. x - y Ú 1 44. 2x + 3y Ú 12
25. - x + y … 4 26. x - y … 3
2x - y Ú 3 - x + 3y … 3
x … 10 x … 9
0 … y … 6 0 … y … 5
x Ú 0 x Ú 0
y Ú 0 y Ú 0 45. x - y Ú 0 46. x + y … 10
x + y … 5 5x + 3y Ú 15
27. 2x + 5y Ú 12 28. - x + 2y … 2
x + 2y … 6 - 2x + 3y … 15
x + 4y … 12 2x - y … 2
y Ú 1 2x - 5y … 6
x Ú 0 x Ú 0
y Ú 0 y Ú 0 47. 16x + 13y … 120 48. 2x + 2y … 21
3x + 4y Ú 25 - 10x + 5y … 24
- 4x + 3y … 11 3x + 5y Ú 37
Problems 49 and 50 introduce an algebraic process for finding the production schedules that will result in a daily profit of
corner points of a solution region without drawing a graph. We $1,100? How are these schedules related to the graph of
will discuss this process later in the chapter. the line 50x + 60y = 1,100?
49. Consider the following system of inequalities and corre- (B) Find a production schedule that will produce a daily
sponding boundary lines: profit greater than $1,100 and repeat part (A) for this
schedule.
3x + 4y … 36 3x + 4y = 36
3x + 2y … 30 3x + 2y = 30 (C) Discuss methods for using lines like those in parts (A)
and (B) to find the largest possible daily profit.
x Ú 0 x = 0
y Ú 0 y = 0 54. Furniture. Refer to Problem 52. The company makes a profit
(A) Use algebraic methods to find the intersection points of $50 on each table and a profit of $15 on each chair.
(if any exist) for each possible pair of boundary lines. (A) If the company makes 20 tables and 20 chairs per day,
(There are six different possible pairs.) the daily profit will be $1,300. Are there other produc-
(B) Test each intersection point in all four inequalities to tion schedules that will result in a daily profit of $1,300?
determine which are corner points. How are these schedules related to the graph of the line
50x + 15y = 1,300?
50. Repeat Problem 49 for
(B) Find a production schedule that will produce a daily
2x + y … 16 2x + y = 16
profit greater than $1,300 and repeat part (A) for this
2x + 3y … 36 2x + 3y = 36 schedule.
x Ú 0 x = 0
(C) Discuss methods for using lines like those in parts (A)
y Ú 0 y = 0 and (B) to find the largest possible daily profit.
55. Plant food. A farmer can buy two types of plant food, mix
A and mix B. Each cubic yard of mix A contains 20 pounds
Applications
of phosphoric acid, 30 pounds of nitrogen, and 5 pounds
of potash. Each cubic yard of mix B contains 10 pounds of
phosphoric acid, 30 pounds of nitrogen, and 10 pounds of
51. Water skis. A manufacturing company makes two types of potash. The minimum monthly requirements are 460 pounds
water skis, a trick ski and a slalom ski. The trick ski requires of phosphoric acid, 960 pounds of nitrogen, and 220 pounds
6 labor-hours for fabricating and 1 labor-hour for finishing. of potash. If x is the number of cubic yards of mix A used and
The slalom ski requires 4 labor-hours for fabricating and y is the number of cubic yards of mix B used, write a system
1 labor-hour for finishing. The maximum labor-hours avail- of linear inequalities that indicates appropriate restraints on
able per day for fabricating and finishing are 108 and 24, x and y. Find the set of feasible solutions graphically for the
respectively. If x is the number of trick skis and y is the num- amounts of mix A and mix B that can be used.
ber of slalom skis produced per day, write a system of linear
inequalities that indicates appropriate restraints on x and y. 56. Nutrition. A dietitian in a hospital is to arrange a special diet
Find the set of feasible solutions graphically for the number using two foods. Each ounce of food M contains 30 units of
of each type of ski that can be produced. calcium, 10 units of iron, and 10 units of vitamin A. Each
ounce of food N contains 10 units of calcium, 10 units of
52. Furniture. A furniture manufacturing company manufactures iron, and 30 units of vitamin A. The minimum requirements
dining-room tables and chairs. A table requires 8 labor- in the diet are 360 units of calcium, 160 units of iron, and
hours for assembling and 2 labor-hours for finishing. A chair 240 units of vitamin A. If x is the number of ounces of food
requires 2 labor-hours for assembling and 1 labor-hour for M used and y is the number of ounces of food N used, write
finishing. The maximum labor-hours available per day for a system of linear inequalities that reflects the conditions
assembly and finishing are 400 and 120, respectively. If x is indicated. Find the set of feasible solutions graphically for
the number of tables and y is the number of chairs produced the amount of each kind of food that can be used.
per day, write a system of linear inequalities that indicates
appropriate restraints on x and y. Find the set of feasible solu- 57. Psychology. A psychologist uses two types of boxes when
tions graphically for the number of tables and chairs that can studying mice and rats. Each mouse spends 10 minutes per
be produced. day in box A and 20 minutes per day in box B. Each rat
spends 20 minutes per day in box A and 10 minutes per day
53. Water skis. Refer to Problem 51. The company makes a
in box B. The total maximum time available per day is 800
profit of $50 on each trick ski and a profit of $60 on each
minutes for box A and 640 minutes for box B. If x is the
slalom ski.
number of mice used and y the number of rats used, write a
(A) If the company makes 10 trick skis and 10 slalom skis system of linear inequalities that indicates appropriate restric-
per day, the daily profit will be $1,100. Are there other tions on x and y. Find the set of feasible solutions graphically.
Answers to Matched Problems 3. (A)

Maximum
1. y Labor-Hours
Available
Labor-Hours Required per Month
20 Two-Person Four-
Solution Boat Person Boat
region 10 (6, 3)
Cutting
Department 0.9 1.8 864
x
220 10 20 Assembly
210 Department 0.8 1.2 672
x 2 2y 5 0 3x 1 y 5 21
(B) 0.9x + 1.8y … 864
2. y 0.8x + 1.2y … 672
x Ú 0
y Ú 0
20 (0, 20) Solution
region y
10 (2, 10)
(9, 3)
(18, 0)
x
0 10 20 400
(480, 240)
210 x 1 3y 5 18 Feasible 0.9x 1 1.8y 5 864
region
5x 1 y 5 20 x 1 y 5 12 x
0 500
0.8x 1 1.2y 5 672
4.3 Linear Programming in Two Dimensions:

A Geometric Approach
■■ A Linear Programming Problem Several problems discussed in the preceding section are related to a more general
■■ General Description of Linear
type of problem called a linear programming problem. Linear programming is a
Programming mathematical process that has been developed to help management in decision mak-
ing. We introduce this topic by considering an example in detail, using an intuitive
■■ Geometric Method for Solving geometric approach. Insight gained from this approach will prove invaluable when
Linear Programming Problems later we consider an algebraic approach that is less intuitive but necessary to solve
■■ Applications most real-world problems.
A Linear Programming Problem

We begin our discussion with a concrete example. The solution method will suggest
two important theorems and a simple general geometric procedure for solving linear
programming problems in two variables.
EXAMPLE 1 Production Scheduling A manufacturer of lightweight mountain tents makes a

standard model and an expedition model. Each standard tent requires 1 labor-hour
from the cutting department and 3 labor-hours from the assembly department. Each
expedition tent requires 2 labor-hours from the cutting department and 4 labor-
hours from the assembly department. The maximum labor-hours available per day
in the cutting and assembly departments are 32 and 84, respectively. If the company
makes a profit of $50 on each standard tent and $80 on each expedition tent, how
many tents of each type should be manufactured each day to maximize the total
daily profit (assuming that all tents can be sold)?
SOLUTION This is an example of a linear programming problem. We begin by ana-
lyzing the question posed in this example.
SECTION 4.3 Linear Programming in Two Dimensions: A Geometric Approach 251
According to the question, the objective of management is to maximize profit.

Since the profits for standard and expedition tents differ, management must decide
how many of each type of tent to manufacture. So it is reasonable to introduce the
following decision variables:
Let x = number of standard tents produced per day
y = number of expedition tents produced per day
Now we summarize the manufacturing requirements, objectives, and restrictions in
Table 1, with the decision variables related to the columns in the table.
Table 1
Labor-Hours per Tent Maximum Labor-Hours
Available per Day
Standard Model Expedition Model
Cutting
department 1 2 32
Assembly
department 3 4 84
Profit per tent $50 $80
Using the last line of Table 1, we form the objective function, in this case the
profit P, in terms of the decision variables (we assume that all tents manufactured
are sold):
P = 50x + 80y Objective function
The objective is to find values of the decision variables that produce the optimal
value (in this case, maximum value) of the objective function.
The form of the objective function indicates that the profit can be made as large
as we like, simply by producing enough tents. But any manufacturing company has
limits imposed by available resources, plant capacity, demand, and so on. These
limits are referred to as problem constraints. Using the information in Table 1, we
can determine two problem constraints.
daily cutting daily cutting maximum labor@ Cutting

£ time for x ≥ + £ time for y ≥ … £ hours available ≥ department
standard tents expedition tents per day constraint
1x + 2y … 32
daily assembly daily assembly maximum labor@ Assembly
£ time for x ≥ + £ time for y ≥ … £ hours available ≥ department
standard tents expedition tents per day constraint
3x + 4y … 84
It is not possible to manufacture a negative number of tents; thus, we have the non-
negative constraints
x Ú 0 and y Ú 0
which we usually write in the form
x, y Ú 0 Nonnegative constraints
We now have a mathematical model for the problem under consideration:
Maximize P = 50x + 80y Objective function
subject to x + 2y … 32
f Problem constraints
3x + 4y … 84
Solving the set of linear inequality constraints graphically, we obtain the feasible
region for production schedules (Fig. 1).
y
20 Assembly
capacity line
(0, 16) 3x 1 4y 5 84
10 (12, 10)
(20, 6) Cutting
Feasible capacity line
region
(23, 2) x 1 2y 5 32
x
(0, 0) 10 20 30 40
(28, 0)
Figure 1
By choosing a production schedule 1x, y2 from the feasible region, a profit can
be determined using the objective function
P = 50x + 80y
For example, if x = 12 and y = 10, the profit for the day would be
P = 501122 + 801102
= $1,400
Or if x = 23 and y = 2, the profit for the day would be
P = 501232 + 80122
= $1,310
Out of all possible production schedules 1x, y2 from the feasible region, which
schedule(s) produces the maximum profit? This is a maximization problem. Since
point-by-point checking is impossible (there are infinitely many points to check),
we must find another way.
By assigning P in P = 50x + 80y a particular value and plotting the resulting
equation in the coordinate system shown in Figure 1, we obtain a constant-profit line.
Every point in the feasible region on this line represents a production schedule that
will produce the same profit. By doing this for a number of values for P, we obtain
a family of constant-profit lines (Fig. 2) that are parallel to each other, since they all
have the same slope. To see this, we write P = 50x + 80y in the slope-intercept form
5 P
y = - x +
8 80
y
$2
,00
Constant-profit line
0
20
P 5 50x 1 80y
with P 5 $2,000
$1
(0, 16)
,48
0
Optimal
$1
,2
10
00
solution
(20, 6)
$8
00
$4
0
x
0
(0, 0) 10 20 30 40
(28, 0)
Figure 2 Constant-profit lines
and note that for any profit P, the constant-profit line has slope - 58. We also observe
that as the profit P increases, the y intercept (P/80) increases, and the line moves
away from the origin.
Therefore, the maximum profit occurs at a point where a constant-profit line
is the farthest from the origin but still in contact with the feasible region, in this
example, at 120, 62 (see Fig. 2). So profit is maximized if the manufacturer makes
20 standard tents and 6 expedition tents per day, and the maximum profit is
P = 501202 + 80162
= $1,480
The point 120, 62 is called an optimal solution to the problem because it maxi-
mizes the objective (profit) function and is in the feasible region. In general, it
appears that a maximum profit occurs at one of the corner points. We also note that
the minimum profit 1P = 02 occurs at the corner point 10, 02.
Matched Problem 1 A manufacturing plant makes two types of inflatable

boats—a two-person boat and a four-person boat. Each two-person boat requires
0.9 labor-hour from the cutting department and 0.8 labor-hour from the assembly
department. Each four-person boat requires 1.8 labor-hours from the cutting de-
partment and 1.2 labor-hours from the assembly department. The maximum labor-
hours available per month in the cutting department and the assembly department
are 864 and 672, respectively. The company makes a profit of $25 on each two-
person boat and $40 on each four-person boat.
(A) Identify the decision variables.
(B) Summarize the relevant material in a table similar to Table 1 in Example 1.
(C) Write the objective function P.
(D) Write the problem constraints and nonnegative constraints.
(E) Graph the feasible region. Include graphs of the objective function for
P = $5,000, P = $10,000, P = $15,000, and P = $21,600.
(F) From the graph and constant-profit lines, determine how many boats should be
manufactured each month to maximize the profit. What is the maximum profit?
Before proceeding further, let’s summarize the steps we used to form the model
in Example 1.
PROCEDURE Constructing a Model for an Applied Linear Programming

Problem
Step 1 Introduce decision variables.
Step 2 Summarize relevant material in table form, relating columns to the decision
variables, if possible (see Table 1).
Step 3 Determine the objective and write a linear objective function.
Step 4 Write problem constraints using linear equations and/or inequalities.
Step 5 Write nonnegative constraints.

Refer to the feasible region S shown in Figure 3.
(A) Let P = x + y. Graph the constant-profit lines through the points 15, 52 and
110, 102. Place a straightedge along the line with the smaller profit and slide
it in the direction of increasing profit, without changing its slope. What is the
maximum value of P? Where does this maximum value occur?
(0, 20)
20
15 (15, 15)
S
10 (10, 10)
5 (5, 5)
(24, 0)
x
(0, 0) 5 10 15 20 25
Figure 3
(B) Repeat part (A) for P = x + 10y.

(C) Repeat part (A) for P = 10x + y.
General Description of Linear Programming

In Example 1 and Matched Problem 1, the optimal solution occurs at a corner point
of the feasible region. Is this always the case? The answer is a qualified yes, as we
will see in Theorem 1. First, we give a few general definitions.
A linear programming problem is one that is concerned with finding the opti-
mal value (maximum or minimum value) of a linear objective function z of the form
z = ax + by, where a and b do not both = 0
and the decision variables x and y are subject to problem constraints in the form
of … or Ú linear inequalities and equations. In addition, the decision variables must
satisfy the nonnegative constraints x Ú 0, y Ú 0. The set of points satisfying both
the problem constraints and the nonnegative constraints is called the feasible region
for the problem. Any point in the feasible region that produces the optimal value of
the objective function over the feasible region is called an optimal solution.
THEOREM 1 Fundamental Theorem of Linear Programming

If the optimal value of the objective function in a linear programming problem
exists, then that value must occur at one or more of the corner points of the feasible
region.
Theorem 1 provides a simple procedure for solving a linear programming prob-

lem, provided that the problem has an optimal solution—not all do. In order to use
Theorem 1, we must know that the problem under consideration has an optimal solu-
tion. Theorem 2 provides some conditions that will ensure that a linear programming
problem has an optimal solution.
THEOREM 2 Existence of Optimal Solutions

(A) If the feasible region for a linear programming problem is bounded, then both the
maximum value and the minimum value of the objective function always exist.
(B) If the feasible region is unbounded and the coefficients of the objective func-
tion are positive, then the minimum value of the objective function exists but
the maximum value does not.
(C) If the feasible region is empty (that is, there are no points that satisfy all the
constraints), then both the maximum value and the minimum value of the
objective function do not exist.
Geometric Method for Solving Linear Programming Problems

The preceding discussion leads to the following procedure for the geometric solution
of linear programming problems with two decision variables:
PROCEDURE Geometric Method for Solving a Linear Programming

Problem with Two Decision Variables
Step 1 Graph the feasible region. Then, if an optimal solution exists according to
Theorem 2, find the coordinates of each corner point.
Step 2 Construct a corner point table listing the value of the objective function at
each corner point.
Step 3 Determine the optimal solution(s) from the table in Step 2.
Step 4 For an applied problem, interpret the optimal solution(s) in terms of the
original problem.
Before we consider more applications, let’s use this procedure to solve some lin-
ear programming problems where the model has already been determined.
EXAMPLE 2 Solving a Linear Programming Problem

(A) Minimize and maximize z = 3x + y
subject to 2x + y … 20
10x + y Ú 36
2x + 5y Ú 36
x, y Ú 0
(B) Minimize and maximize z = 10x + 20y
subject to 6x + 2y Ú 36
2x + 4y Ú 32
y y … 20
x, y Ú 0
(2, 16)
15
SOLUTION
(A) Step 1 Graph the feasible region S (Fig. 4). Then, after checking Theorem 2 to
10
S determine whether an optimal solution exists, find the coordinates of
each corner point. Since S is bounded, z will have both a maximum and
5 (3, 6) (8, 4) a minimum value on S (Theorem 2A) and these will both occur at corner
points (Theorem 1).
x Step 2 Evaluate the objective function at each corner point, as shown in the
0 5 10 table.
Figure 4
Corner Point
1x, y2 z = 3x + y
13, 62 15
12, 162 22
18, 42 28
Step 3 Determine the optimal solutions from Step 2. Examining the values
in the table, we see that the minimum value of z is 15 at (3, 6) and the
maximum value of z is 28 at (8, 4).
(B) Step 1 Graph the feasible region S (Fig. 5). Then, after checking Theorem 2
to determine whether an optimal solution exists, find the coordinates
of each corner point. Since S is unbounded and the coefficients of the
y objective function are positive, z has a minimum value on S but no

(0, 20) maximum value (Theorem 2B).
20
Step 2 Evaluate the objective function at each corner point, as shown in the
(0, 18)
table.
15
S Corner Point
1x, y2 z = 10x + 20y
10
10, 202 400
(4, 6) 10, 182 360
5
14, 62 160
(16, 0) 116, 02 160

x
0 5 10 15 20
Step 3 Determine the optimal solution from Step 2. The minimum value of z is
Figure 5 160 at 14, 62 and at 116, 02.
The solution to Example 2B is a multiple optimal solution. In general, if two
corner points are both optimal solutions to a linear programming problem, then any
point on the line segment joining them is also an optimal solution. This is the only
way that optimal solutions can occur at noncorner points.
Matched Problem 2
(A) Maximize and minimize z = 4x + 2y subject to the constraints given in
Example 2A.
(B) Maximize and minimize z = 20x + 5y subject to the constraints given in
Example 2B.
CONCEPTUAL INSIGHT
Determining that an optimal solution exists is a critical step in the solution of a
linear programming problem. If you skip this step, you may examine a corner
point table like the one in the solution of Example 2B and erroneously conclude
that the maximum value of the objective function is 400.

In Example 2B we saw that there was no optimal solution for the problem of
maximizing the objective function z over the feasible region S. We want to add an
additional constraint to modify the feasible region so that an optimal solution for the
maximization problem does exist. Which of the following constraints will accom-
plish this objective?
(A) x … 20 (B) y Ú 4 (C) x … y (D) y … x
y For an illustration of Theorem 2C, consider the following:
10 Maximize P = 2x + 3y
subject to x + y Ú 8
B
x + 2y … 8
5 2x + y … 10
x, y Ú 0
A
x The intersection of the graphs of the constraint inequalities is the empty set (Fig. 6); so
0 5 10 the feasible region is empty. If this happens, the problem should be reexamined to see
A>B5[ if it has been formulated properly. If it has, the management may have to reconsider
Figure 6 items such as labor-hours, overtime, budget, and supplies allocated to the project in
order to obtain a nonempty feasible region and a solution to the original problem.
Applications
EXAMPLE 3 Medication A hospital patient is required to have at least 84 units of drug A and
120 units of drug B each day (assume that an overdose of either drug is harmless).
Each gram of substance M contains 10 units of drug A and 8 units of drug B, and
each gram of substance N contains 2 units of drug A and 4 units of drug B. Now,
suppose that both M and N contain an undesirable drug D: 3 units per gram in M and
1 unit per gram in N. How many grams of each of substances M and N should be
mixed to meet the minimum daily requirements and simultaneously minimize the
intake of drug D? How many units of the undesirable drug D will be in this mixture?
SOLUTION First we construct the mathematical model.
Step 1 Introduce decision variables. According to the questions asked, we must
decide how many grams of substances M and N should be mixed to form the
daily dose of medication. These two quantities are the decision variables:
x = number of grams of substance M used
y = number of grams of substance N used
Step 2 Summarize relevant material in a table, relating the columns to substances
M and N.
Minimum Daily
Amount of Drug per Gram Requirement
Substance M Substance N
Drug A 10 units/gram 2 units/gram 84 units
Drug B 8 units/gram 4 units/gram 120 units
Drug D 3 units/gram 1 unit/gram
Step 3 Determine the objective and the objective function. The objective is to mini-
mize the amount of drug D in the daily dose of medication. Using the decision
variables and the information in the table, we form the linear objective function
C = 3x + y
Step 4 Write the problem constraints. The constraints in this problem involve mini-
mum requirements, so the inequalities will take a different form:
10x + 2y Ú 84 Drug A constraint
y
8x + 4y Ú 120 Drug B constraint
40
(0, 42) Step 5 Add the nonnegative constraints and summarize the model.
30
Minimize C = 3x + y Objective function
Feasible subject to 10x + 2y Ú 84 Drug A constraint
(4, 22) region
20 8x + 4y Ú 120 Drug B constraint
10
(15, 0)
Now we use the geometric method to solve the problem.
0
x Step 1 Graph the feasible region (Fig. 7). Then, after checking Theorem 2 to deter-
10 20 30 40
mine whether an optimal solution exists, find the coordinates of each corner
Figure 7
point. Since the feasible region is unbounded and the coefficients of the objec-
Corner Point tive function are positive, this minimization problem has a solution.
1x, y2 C = 3x + y Step 2 Evaluate the objective function at each corner point, as shown in the table.
Step 3 Determine the optimal solution from Step 2. The optimal solution is C = 34
10, 422 42
at the corner point 14, 222.
14, 222 34
Step 4 Interpret the optimal solution in terms of the original problem. If we use
115, 02 45
4 grams of substance M and 22 grams of substance N, we will supply the
minimum daily requirements for drugs A and B and minimize the intake of
the undesirable drug D at 34 units. (Any other combination of M and N from
the feasible region will result in a larger amount of the undesirable drug D.)
Matched Problem 3 A chicken farmer can buy a special food mix A at 20¢
per pound and a special food mix B at 40¢ per pound. Each pound of mix A con-
tains 3,000 units of nutrient N1 and 1,000 units of nutrient N2; each pound of mix B
contains 4,000 units of nutrient N1 and 4,000 units of nutrient N2. If the minimum
daily requirements for the chickens collectively are 36,000 units of nutrient N1 and
20,000 units of nutrient N2, how many pounds of each food mix should be used
each day to minimize daily food costs while meeting (or exceeding) the minimum
daily nutrient requirements? What is the minimum daily cost? Construct a math-
ematical model and solve using the geometric method.
CONCEPTUAL INSIGHT
Refer to Example 3. If we change the minimum requirement for drug B from 120
to 125, the optimal solution changes to 3.6 grams of substance M and 24.1 grams
of substance N, correct to one decimal place.
Now refer to Example 1. If we change the maximum labor-hours available
per day in the assembly department from 84 to 79, the solution changes to 15
standard tents and 8.5 expedition tents.
We can measure 3.6 grams of substance M and 24.1 grams of substance
N, but how can we make 8.5 tents? Should we make 8 tents? Or 9 tents? If the
solutions to a problem must be integers and the optimal solution found graphi-
cally involves decimals, then rounding the decimal value to the nearest integer
does not always produce the optimal integer solution (see Problem 44, Exercises
4.3). Finding optimal integer solutions to a linear programming problem is called
integer programming and requires special techniques that are beyond the scope
of this book. As mentioned earlier, if we encounter a solution like 8.5 tents per
day, we will interpret this as an average value over many days of production.
Exercises 4.3
W Skills Warm-up Exercises A In Problems 9–12, graph the constant-profit lines through 13, 32
and 16, 62. Use a straightedge to identify the corner point where
In Problems 1–8, if necessary, review Theorem 1. In Problems
the maximum profit occurs (see Explore and Discuss 1). Confirm
1–4, the feasible region is the set of points on and inside the rect-
your answer by constructing a corner-point table.
angle with vertices 10, 02, 112, 02, 10, 52, and 112, 52. Find the
y
maximum and minimum values of the objective function Q over
the feasible region. 15
1. Q = 7x + 14y 2. Q = 3x + 15y (0, 12)
3. Q = 10x - 12y 4. Q = - 9x + 20y 10 (7, 9)
S
In Problems 5–8, the feasible region is the set of points on and
inside the triangle with vertices 10, 02, 18, 02, and 10, 102. 5 (6, 6)
Find the maximum and minimum values of the objective function (3, 3)
Q over the feasible region. x
(0, 0) 5 (10, 0) 15
5. Q = - 4x - 3y 6. Q = 3x + 2y
9. P = x + y 10. P = 4x + y
7. Q = - 6x + 4y 8. Q = 10x - 8y
11. P = 3x + 7y 12. P = 9x + 3y
In Problems 13–16, graph the constant-cost lines through 19, 92 24. Minimize and maximize
and 112, 122. Use a straightedge to identify the corner point z = 8x + 7y
where the minimum cost occurs. Confirm your answer by con- subject to 4x + 3y Ú 24
structing a corner-point table. 3x + 4y Ú 8
y x, y Ú 0
(15, 15) 25. Minimize and maximize
(0, 15)
z = 50x + 40y
(12, 12)
T subject to 3x + 8y Ú 24
10 x + y Ú 5
(0, 8) (9, 9) 3x + 2y Ú 12
5 x, y Ú 0
26. Maximize P = 20x + 10y
(4, 3)
subject to 3x + y … 21
x
0 5 10 (15, 0) x + y … 9
13. C = 7x + 4y 14. C = 7x + 9y x + 3y … 21
x, y Ú 0
15. C = 3x + 8y 16. C = 2x + 11y
B Solve the linear programming problems stated in Problems 17–38. subject to x - y … -4
x + y … 8
x - y … 4
subject to 3x + 2y … 12
x, y Ú 0
x, y Ú 0
28. Minimize and maximize
z = 400x + 100y
18. Maximize P = 30x + 12y subject to 3x + y Ú 24
subject to 3x + y … 18 x + y Ú 16
x, y Ú 0 x + 3y Ú 30
x, y Ú 0
19. Minimize C = 5x + 7y
P = 30x + 10y
subject to x + 2y Ú 10
x, y Ú 0
6x + 4y … 36
2x + y … 10
20. Minimize C = 15x + 25y x, y Ú 0
subject to 4x + 7y Ú 28 30. Minimize and maximize
x, y Ú 0 P = 2x + y
21. Maximize P = 20x + 10y 6x + 4y … 36
subject to 5x + 3y … 42 4x + 2y … 20
3x + y … 24 x, y Ú 0
x, y Ú 0 31. Minimize and maximize
22. Maximize P = 3x + 2y P = 3x + 5y
subject to 6x + 3y … 24
x + y … 4
3x + 6y … 30
2x + 3y Ú 12
x, y Ú 0
x, y Ú 0
z = 5x + 2y 32. Minimize and maximize
subject to x + y Ú 5 P = - x + 3y
x + 2y Ú 6 subject to 2x - y Ú 4
x, y Ú 0 - x + 2y … 4
y … 6
x, y Ú 0
33. Minimize and maximize Problems 41–48 refer to the bounded feasible region with corner
P = 6x + 3y points O = 10, 02, A = 10, 52, B = 14, 32, and C = 15, 02
subject to 4x + y Ú 80 that is determined by the system of inequalities
x + 5y Ú 115 x + 2y … 10
3x + 2y … 150
3x + y … 15
x, y Ú 0
x, y Ú 0
41. If P = ax + 10y, find all numbers a such that the maximum
P = 12x + 14y
value of P occurs only at B.
subject to - 2x + y Ú 6
x + y … 15 42. If P = ax + 10y, find all numbers a such that the maximum
3x - y Ú 0 value of P occurs only at A.
x, y Ú 0 43. If P = ax + 10y, find all numbers a such that the maximum
value of P occurs only at C.
subject to 0.8x + 0.1y … 160 44. If P = ax + 10y, find all numbers a such that the maximum
value of P occurs at both A and B.
2.8x + 0.4y … 560
0.04x + 0.02y … 20 45. If P = ax + 10y, find all numbers a such that the maximum
x, y Ú 0 value of P occurs at both B and C.
46. If P = ax + 10y, find all numbers a such that the minimum
36. Minimize C = 30x + 10y value of P occurs only at C.
subject to 1.8x + 0.9y Ú 270
47. If P = ax + 10y, find all numbers a such that the minimum
0.3x + 0.2y Ú 54
value of P occurs at both O and C.
0.01x + 0.03y Ú 3.9
x, y Ú 0 48. If P = ax + 10y, explain why the minimum value of P
cannot occur at B.
37. Maximize P = 525x + 478y
subject to 275x + 322y … 3,381
350x + 340y … 3,762
425x + 306y … 4,114 Applications
x, y Ú 0
In Problems 49–64, construct a mathematical model in the form
38. Maximize P = 300x + 460y of a linear programming problem. (The answers in the back of
subject to 245x + 452y … 4,181 the book for these application problems include the model.) Then
solve by the geometric method.
290x + 379y … 3,888
390x + 299y … 4,407 49. Water skis. A manufacturing company makes two types of
x, y Ú 0 water skis—a trick ski and a slalom ski. The relevant manu-
facturing data are given in the table below.
C In Problems 39 and 40, explain why Theorem 2 cannot be used to
conclude that a maximum or minimum value exists. Graph the fea- Labor-Hours per Ski Maximum Labor-Hours
sible regions and use graphs of the objective function z = x - y Department Trick Ski Slalom Ski Available per Day
for various values of z to discuss the existence of a maximum Fabricating 6 4 108
value and a minimum value. Finishing 1 1 24

(A) If the profit on a trick ski is $40 and the profit on a
z = x - y slalom ski is $30, how many of each type of ski should
subject to x - 2y … 0 be manufactured each day to realize a maximum profit?
2x - y … 6 What is the maximum profit?
x, y Ú 0 (B) Discuss the effect on the production schedule and the
40. Minimize and maximize maximum profit if the profit on a slalom ski decreases
z = x - y to $25.
subject to x - 2y Ú - 6 (C) Discuss the effect on the production schedule and the maxi-
2x - y Ú 0 mum profit if the profit on a slalom ski increases to $45.
x, y Ú 0
50. Furniture. A furniture manufacturing company manufactures
dining-room tables and chairs. The relevant manufacturing
data are given in the table below.
Labor-Hours per Unit Maximum Labor-Hours rent in order to minimize the transportation costs? What are
Department Table Chair Available per Day the minimal transportation costs?
Assembly 8 2 400 54. Transportation. Refer to Problem 53. If each van can trans-
Finishing 2 1 120 port 7 people and there are 35 available chaperones, show
Profit per unit $90 $25 that the optimal solution found graphically involves decimals.
Find all feasible solutions with integer coordinates and iden-
(A) How many tables and chairs should be manufactured tify the one that minimizes the transportation costs. Can this
each day to realize a maximum profit? What is the maxi- optimal integer solution be obtained by rounding the optimal
mum profit? decimal solution? Explain.
55. Investment. An investor has $60,000 to invest in a CD and a
(B) Discuss the effect on the production schedule and the
mutual fund. The CD yields 5% and the mutual fund yields an
maximum profit if the marketing department of the com-
average of 9%. The mutual fund requires a minimum invest-
pany decides that the number of chairs produced should
ment of $10,000, and the investor requires that at least twice
be at least four times the number of tables produced.
as much should be invested in CDs as in the mutual fund. How
51. Production scheduling. A furniture company has two plants much should be invested in CDs and how much in the mutual
that produce the lumber used in manufacturing tables and fund to maximize the return? What is the maximum return?
chairs. In 1 day of operation, plant A can produce the lumber
56. Investment. An investor has $24,000 to invest in bonds of
required to manufacture 20 tables and 60 chairs, and plant
AAA and B qualities. The AAA bonds yield an average of 6%,
B can produce the lumber required to manufacture 25 tables
and the B bonds yield 10%. The investor requires that at least
and 50 chairs. The company needs enough lumber to manu-
three times as much money should be invested in AAA bonds
facture at least 200 tables and 500 chairs.
as in B bonds. How much should be invested in each type of
(A) If it costs $1,000 to operate plant A for 1 day and $900 bond to maximize the return? What is the maximum return?
to operate plant B for 1 day, how many days should each
57. Pollution control. Because of new federal regulations on
plant be operated to produce a sufficient amount of lum-
pollution, a chemical plant introduced a new, more expensive
ber at a minimum cost? What is the minimum cost?
process to supplement or replace an older process used in
(B) Discuss the effect on the operating schedule and the mini- the production of a particular chemical. The older process
mum cost if the daily cost of operating plant A is reduced emitted 20 grams of sulfur dioxide and 40 grams of particu-
to $600 and all other data in part (A) remain the same. late matter into the atmosphere for each gallon of chemical
produced. The new process emits 5 grams of sulfur dioxide
(C) Discuss the effect on the operating schedule and the mini-
and 20 grams of particulate matter for each gallon produced.
mum cost if the daily cost of operating plant B is reduced
The company makes a profit of 60¢ per gallon and 20¢ per
to $800 and all other data in part (A) remain the same.
gallon on the old and new processes, respectively.
52. Mobiles. A mobile firm manufactures two types of
(A) If the government allows the plant to emit no more than
mobiles: a dial pad model and a touch screen model. The pro-
16,000 grams of sulfur dioxide and 30,000 grams of par-
duction of a dial pad mobile requires a capital expenditure of
ticulate matter daily, how many gallons of the chemical
$100 and 8 hours of labor. The production of a touch screen
should be produced by each process to maximize daily
mobile requires a capital expenditure of $50 and 24 hours of
profit? What is the maximum daily profit?
labor. The firm has a capital of $5,000 and 5,000 labor hours
available for the production of dial pad and touch screen (B) Discuss the effect on the production schedule and the
mobiles. maximum profit if the government decides to restrict
emissions of sulfur dioxide to 11,500 grams daily and all
(A) What is the maximum number of mobiles the company
other data remain unchanged.
is capable of manufacturing?
(C) Discuss the effect on the production schedule and the
(B) If each dial pad mobile contributes a profit of $75 and each
maximum profit if the government decides to restrict
touch screen mobile contributes a profit of $25, how much
emissions of sulfur dioxide to 7,200 grams daily and all
profit will the company make by producing the maximum
other data remain unchanged.
number of mobiles determined in part (A)? Is this the
maximum profit? If not, what is the maximum profit? 58. Capital expansion. A fast-food chain plans to expand by
opening several new restaurants. The chain operates two
53. Transportation. The officers of a high school senior class
types of restaurants, drive-through and full-service. A drive-
are planning to rent buses and vans for a class trip. Each bus
through restaurant costs $100,000 to construct, requires
can transport 40 students, requires 3 chaperones, and costs
5 employees, and has an expected annual revenue of $200,000.
$1,200 to rent. Each van can transport 8 students, requires
A full-service restaurant costs $150,000 to construct,
1 chaperone, and costs $100 to rent. Since there are 400
requires 15 employees, and has an expected annual revenue
students in the senior class that may be eligible to go on
of $500,000. The chain has $2,400,000 in capital avail-
the trip, the officers must plan to accommodate at least 400
able for expansion. Labor contracts require that they hire no
students. Since only 36 parents have volunteered to serve as
more than 210 employees, and licensing restrictions require
chaperones, the officers must plan to use at most 36 chap-
that they open no more than 20 new restaurants. How many
erones. How many vehicles of each type should the officers
restaurants of each type should the chain open in order to 63. Psychology. A psychologist uses two types of boxes with
maximize the expected revenue? What is the maximum mice and rats. The amount of time (in minutes) that each
expected revenue? How much of their capital will they use and mouse and each rat spends in each box per day is given in the
how many employees will they hire? table. What is the maximum number of mice and rats that can
be used in this experiment? How many mice and how many
59. Fertilizer. A fruit grower can use two types of fertilizer in his
rats produce this maximum?
orange grove, brand A and brand B. The amounts (in pounds)
of nitrogen, phosphoric acid, and chloride in a bag of each Time Maximum Time
brand are given in the table. Tests indicate that the grove Mice Rats Available per Day
needs at least 1,000 pounds of phosphoric acid and at most Box A 10 min 20 min 800 min
400 pounds of chloride. Box B 20 min 10 min 640 min
Pounds per Bag

64. Music. A music academy conducts a training course in
Brand A Brand B
vocal and guitar for both amateurs and professionals.
Nitrogen 8 3 Allocation of classes and cost of conducting the classes are
Phosphoric acid 4 4 given in the table. How many classes of vocal and guitar
Chloride 2 1 should be arranged to minimize the cost and meet the class-
hour requirements? What is the minimum weekly cost of
conducting classes?
(A) If the grower wants to maximize the amount of nitrogen
added to the grove, how many bags of each mix should Class-Hours Minimum Class-
be used? How much nitrogen will be added? Hours to Be
Vocal Guitar Attended
(B) If the grower wants to minimize the amount of nitrogen Amateurs 2 3 15
added to the grove, how many bags of each mix should
Professionals 1 2 8
be used? How much nitrogen will be added?
Costs per class $7 $12
60. Nutrition. A dietitian is to arrange a special diet composed of
two foods, M and N. Each ounce of food M contains 30 units
of calcium, 10 units of iron, 10 units of vitamin A, and 8 units Answers to Matched Problems
of cholesterol. Each ounce of food N contains 10 units of
calcium, 10 units of iron, 30 units of vitamin A, and 4 units of 1. (A) x = number of two-person boats produced each month
cholesterol. If the minimum daily requirements are 360 units y = number of four-person boats produced each month
of calcium, 160 units of iron, and 240 units of vitamin A, (B)
how many ounces of each food should be used to meet the Labor-Hours Required Maximum Labor-
minimum requirements and at the same time minimize the Two-Person Four-Person Hours Available
cholesterol intake? What is the minimum cholesterol intake? Boat Boat per Month
Cutting
61. Plant food. A farmer can buy two types of plant food, mix
department 0.9 1.8 864
A and mix B. Each cubic yard of mix A contains 20 pounds
of phosphoric acid, 30 pounds of nitrogen, and 5 pounds Assembly
department 0.8 1.2 672
of potash. Each cubic yard of mix B contains 10 pounds of
phosphoric acid, 30 pounds of nitrogen, and 10 pounds of Profit per
boat $25 $40
potash. The minimum monthly requirements are 460 pounds
of phosphoric acid, 960 pounds of nitrogen, and 220 pounds (C) P = 25x + 40y
of potash. If mix A costs $30 per cubic yard and mix B (D) 0.9x + 1.8y … 864
costs $35 per cubic yard, how many cubic yards of each 0.8x + 1.2y … 672
mix should the farmer blend to meet the minimum monthly x, y Ú 0
requirements at a minimal cost? What is this cost?
(E) y
62. Animal food. A laboratory technician in a medical research
center is asked to formulate a diet from two commercially P 5 21,600
packaged foods, food A and food B, for a group of animals.
Each ounce of food A contains 8 units of fat, 16 units of
carbohydrate, and 2 units of protein. Each ounce of food B P 5 15,000
contains 4 units of fat, 32 units of carbohydrate, and 8 units 300
of protein. The minimum daily requirements are 176 units P 5 10,000 (480, 240)
of fat, 1,024 units of carbohydrate, and 384 units of protein. 200
If food A costs 5¢ per ounce and food B costs 5¢ per ounce, P 5 5,000
how many ounces of each food should be used to meet the
minimum daily requirements at the least cost? What is the x
cost for this amount of food? 0 500 1,000
(F) 480 two-person boats, 240 four-person boats; Max 3. Min C = 0.2x + 0.4y
P = $21,600 per month subject to 3,000x + 4,000y Ú 36,000
2. (A) Min z = 24 at 13, 62; Max z = 40 at 12, 162 and 1,000x + 4,000y Ú 20,000
18, 42 (multiple optimal solution) x, y Ú 0
(B) Min z = 90 at 10, 182; no maximum value 8 lb of mix A, 3 lb of mix B; Min C = $2.80 per day

4.1 Linear Inequalities in Two Variables EXAMPLES
• A line divides the plane into two regions called half-planes. A vertical line divides the plane into left
and right half-planes; a nonvertical line divides it into upper and lower half-planes. In either case, the
dividing line is called the boundary line of each half-plane. Ex. 1, p. 238
Ex. 2, p. 238
• The graph of a linear inequality is the half-plane obtained by following the procedure on page 238.
Ex. 3, p. 240
• The variables in an applied problem are often required to be nonnegative. Ex. 4, p. 240
4.2 System of Linear Inequalities in Two Variables
• The solution region (also called the feasible region) of a system of linear inequalities is the graph of all Ex. 1, p. 244
ordered pairs that simultaneously satisfy all the inequalities in the system.
• A corner point of a solution region is a point in the region that is the intersection of two boundary lines. Ex. 2, p. 245
• A solution region is bounded if it can be enclosed in a circle and unbounded if it cannot. Ex. 3, p. 246
4.3 Linear Programming in Two Dimensions: A Geometric Approach

• The problem of finding the optimal (maximum or minimum) value of a linear objective function on a Ex. 1, p. 250
feasible region is called a linear programming problem.
• The optimal value (if it exists) of the objective function in a linear programming problem must occur at Ex. 2, p. 255
one (or more) of the corner points of the feasible region (Theorem 1, page 254). Existence criteria are Ex. 3, p. 257
described in Theorem 2, page 254, and a solution procedure is listed on page 255.
Review Exercises
Work through all the problems in this chapter review and check B In Exercises 7 and 8, state the linear inequality whose graph is
answers in the back of the book. Answers to all review problems given in the figure. Write the boundary line equation in the form
are there, and following each answer is a number in italics indicat- Ax + By = C, with A, B, and C integers, before stating the
ing the section in which that type of problem is discussed. Where inequality.
weaknesses show up, review appropriate sections in the text. 7. y
8. y
A Graph each inequality. 5
10
1. 2y 6 4x + 3 2. 4y - 8x … 20
Graph the systems in Problems 3–6 and indicate whether each
solution region is bounded or unbounded. Find the coordinates of x
0
5
x
25 0 10
each corner point. 210
3. 5x + 9y … 90 4. 15x + 16y Ú 1,200

x, y Ú 0 x, y Ú 0 25
210
5. 3x + y … 7 6. 3x + y Ú 9
Solve the linear programming problems in Problems 9–13.
2x + 8y … 26 2x + 4y Ú 16
x, y Ú 0 x, y Ú 0 9. Maximize P = 3x + 2y
x + 2y … 9
x, y Ú 0
10. Maximize P = 2x + 5y In Problems 15 and 16, construct a mathematical model in the

subject to 2x + y Ú 12 form of a linear programming problem. (The answers in the back
x + 2y … 18 of the book for these application problems include the model.)
Then solve the problem by the indicated method.
x, y Ú 0
15. Sail manufacture. South Shore Sail Loft manufactures
C 11. Maximize P = 3x + 4y regular and competition sails. Each regular sail takes 2 hours
subject to x + 2y … 12 to cut and 4 hours to sew. Each competition sail takes 3 hours
x + y … 7 to cut and 10 hours to sew. There are 150 hours available in
2x + y … 10 the cutting department and 380 hours available in the sewing
x, y Ú 0 department.
12. Minimize C = 8x + 3y (A) If the Loft makes a profit of $100 on each regular sail
and $200 on each competition sail, how many sails of
each type should the company manufacture to maximize
2x + y Ú 15 its profit? What is the maximum profit?
x Ú 3
(B) An increase in the demand for competition sails causes
x, y Ú 0
the profit on a competition sail to rise to $260. Discuss
13. Maximize P = 4x + 3y the effect of this change on the number of sails manufac-
subject to 2x + y Ú 12 tured and on the maximum profit.
x + y Ú 8 (C) A decrease in the demand for competition sails causes
x … 12 the profit on a competition sail to drop to $140. Discuss
y … 12 the effect of this change on the number of sails manufac-
tured and on the maximum profit.
x, y Ú 0
16. Animal food. A special diet for laboratory animals is to con-
tain at least 850 units of vitamins, 800 units of minerals, and
Applications 1,150 calories. There are two feed mixes available, mix A and
mix B. A gram of mix A contains 2 units of vitamins, 2 units
14. Electronics. A company uses two machines to solder circuit of minerals, and 4 calories. A gram of mix B contains 5 units
boards, an oven and a wave soldering machine. A circuit of vitamins, 4 units of minerals, and 5 calories.
board for a calculator needs 4 minutes in the oven and 2 min- (A) If mix A costs $0.04 per gram and mix B costs $0.09
utes on the wave machine, while a circuit board for a toaster per gram, how many grams of each mix should be used
requires 3 minutes in the oven and 1 minute on the wave to satisfy the requirements of the diet at minimal cost?
machine. (Source: Universal Electronics) What is the minimum cost?
(A) How many circuit boards for calculators and toasters can
(B) If the price of mix B decreases to $0.06 per gram, dis-
be produced if the oven is available for 5 hours? Express
cuss the effect of this change on the solution in part (A).
your answer as a linear inequality with appropriate non-
negative restrictions and draw its graph. (C) If the price of mix B increases to $0.12 per gram, discuss
the effect of this change on the solution in part (A).
(B) How many circuit boards for calculators and toasters
can be produced if the wave machine is available for
2 hours? Express your answer as a linear inequality with
appropriate nonnegative restrictions and draw its graph.
5 Linear Programming:
The Simplex Method
5.1 The Table Method:
An Introduction to the
Introduction
Simplex Method The geometric method of solving linear programming problems (presented in
5.2 The Simplex Method: Chapter 4) provides an overview of linear programming. But, practically speaking,
Maximization with the geometric method is useful only for problems involving two decision variables
Problem Constraints of and relatively few problem constraints. What happens when we need more deci-
the Form … sion variables and more problem constraints? We use an algebraic method called
5.3 The Dual Problem: the simplex method, developed by George B. Dantzig (1914–2005) in 1947.
Minimization with Ideally suited to computer use, the simplex method is used routinely on applied
Problem Constraints of problems involving thousands of variables and problem constraints. In this chapter,
the Form Ú we move from an introduction to the simplex method to the big M method, which
5.4 Maximization and can be used to solve linear programming problems with a large number of vari-
Minimization with Mixed ables and constraints. We also explore many applications. An oil refining com-
Problem Constraints pany, for example, can use the simplex method to schedule its operations so that
production quotas are met at minimum cost (see Problem 50 in Section 5.3).
265
266 CHAPTER 5 Linear Programming: The Simplex Method
5.1 The Table Method: An Introduction to the Simplex Method

■■ Standard Maximization Problems in In Chapter 4, we denoted variables by single letters such as x and y. In this chapter,
Standard Form we will use letters with subscripts, for example, x1, x2, x3, to denote variables. With
■■ Slack Variables this new notation, we will lay the groundwork for solving linear programming prob-
lems algebraically, by means of the simplex method. In this section, we introduce the
■■ The Table Method: Basic Solutions table method to provide an introduction to the simplex method. Both methods, the
and Basic Feasible Solutions table method and the simplex method, solve linear programming problems without
■■ Basic and Nonbasic Variables the necessity of drawing a graph of the feasible region.
■■ Summary
Standard Maximization Problems in Standard Form
The tent production problem that we considered in Section 4.3 is an example of a
standard maximization problem in standard form. We restate the tent production
problem below using subscript notation for the variables.
Maximize P = 50x1 + 80x2 Objective function
subject to x1 + 2x2 … 32 Cutting department constraint
(1)
3x1 + 4x2 … 84 Assembly department constraint
x1, x2 Ú 0 Nonnegative constraints
The decision variables x1 and x2 are the number of standard and expedition tents,
respectively, produced each day.
Notice that the problem constraints involve … inequalities with nonnegative con-
stants to the right of the inequality. Maximization problems that satisfy this condition
are called standard maximization problems. In this and the next section, we restrict
our attention to standard maximization problems.
DEFINITION Standard Maximization Problem in Standard Form

A linear programming problem is said to be a standard maximization problem
in standard form if its mathematical model is of the following form:
Maximize the objective function
P = c1 x1 + c2 x2 + g + ck xk
subject to problem constraints of the form
a1x1 + a2 x2 + g + ak xk … b b Ú 0
with nonnegative constraints
x1, x2, c , xk Ú 0
Note: Mathematical model (1) is a standard maximization problem in standard
form. The coefficients of the objective function can be any real numbers.

Find an example of a standard maximization problem in standard form involving two
variables and one problem constraint such that
(A) The feasible region is bounded.
(B) The feasible region is unbounded.
Is it possible for a standard maximization problem to have no solution? Explain.
SECTION 5.1 The Table Method: An Introduction to the Simplex Method 267
Slack Variables
To adapt a linear programming problem to the matrix methods used in the simplex
process (as discussed in the next section), we convert the problem constraint inequal-
ities into a system of linear equations using slack variables. In particular, to convert
the system of inequalities from model (1),
x1 + 2x2 … 32 Cutting department constraint
3x1 + 4x2 … 84 Assembly department constraint (2)
into a system of equations, we add variables s1 and s2 to the left sides of the problem
constraint inequalities in (2) to obtain
x1 + 2x2 + s1 = 32
(3)
3x1 + 4x2 + s2 = 84
The variables s1 and s2 are called slack variables because each makes up the
difference (takes up the slack) between the left and right sides of an inequality in
system (2). For example, if we produced 20 standard tents 1x1 = 202 and 5 expedi-
tion tents 1x2 = 52, then the number of labor-hours used in the cutting department
would be 20 + 2152 = 30, leaving a slack of 2 unused labor-hours out of the
32 available. So s1 would be equal to 2.
Notice that if the decision variables x1 and x2 satisfy the system of constraint in-
equalities (2), then the slack variables s1 and s2 are nonnegative.
The Table Method: Basic Solutions and Basic Feasible Solutions

We call the system of inequalities of a linear programming problem an i-system
(“i” for inequality), and we call the associated system of linear equations, obtained
via slack variables, an e-system (“e” for equation).
The solutions of the i-system (2) are the points in the feasible region of Figure 1
(the feasible region was graphed earlier in Section 4.3). For example, 120, 42 is a
solution of the i-system, but 120, 82 is not (Fig. 1).
x2
D(0, 21)
20
A(0, 16) 3x1 1 4x2 5 84
10
B(20, 6)
Feasible region x1 1 2x2 5 32
O(0, 0) E(32, 0)
x1
0
Figure 1 10 C(28, 0) 30 40
The solutions of the e-system (3) are quadruples 1x1, x2, s1, s2 2. Any point
P = 1x1, x2 2 in the plane corresponds to a unique solution of the e-system (3). For
example, 120, 42 corresponds to 120, 4, 4, 82; we say that 120, 4, 4, 82 are the
expanded coordinates of 120, 42. Similarly, the expanded coordinates of 120, 82
are 120, 8, - 4, - 82. Both 120, 4, 4, 82 and 120, 8, - 4, - 82 are solutions of (3).
Systems (2) and (3) do not have the same solutions. However, the solutions of
the i-system (2) correspond to the solutions of the e-system (3) in which the vari-
ables x1, x2, s1, and s2 are all nonnegative. Such a solution of the e-system, in which
the values of all decision variables and slack variables are nonnegative, is called a
feasible solution. So 120, 4, 4, 82 is a feasible solution of the e-system (3), but the
solution 120, 8, - 4, - 82 is not feasible. Note that 120, 42 is in the feasible region
(Fig. 1), but 120, 82 is not. To summarize, the feasible solutions of (3) correspond to
the points of the feasible region shown in Figure 1.
Any line segment that forms a boundary for the feasible region of (2) lies on one
of the following four lines [replace “… ” by “ = ” in each inequality of i-system (2)]:
x1 + 2x2 = 32
3x1 + 4x2 = 84
x1 = 0
x2 = 0
So any corner point of the feasible region must lie on at least two of the four lines.
If P is the intersection of the first two lines, then s1 = 0 and s2 = 0 in the expanded
coordinates for P. If P is the intersection of the first and third lines, then s1 = 0 and
x1 = 0 in the expanded coordinates for P. Continuing in this way, we conclude that
for any corner point P of the feasible region, at least two of the variables x1, x2, s1,
and s2 must equal 0 in the expanded coordinates of P.
The reasoning above gives a procedure, called the table method, for solving a
standard maximization problem in standard form.
PROCEDURE The Table Method (Two Decision Variables)

Assume that a standard maximization problem in standard form has two decision
variables x1 and x2 and m problem constraints.
Step 1 Use slack variables s1, s2, . . ., sm to convert the i-system to an e-system.
Step 2 Form a table with 1m + 221m + 12 >2 rows and m + 2 columns labeled
x1, x2, s1, s2, . . ., sm. In the first row, assign 0 to x1 and x2. In the second
row, assign 0 to x1 and s1. Continue until the rows contain all possible com-
binations of assigning two 0’s to the variables.
Step 3 Complete each row to a solution of the e-system, if possible. Because two
of the variables have the value 0, this involves solving a system of m linear
equations in m variables. Use the Gauss–Jordan method, or another method
if you find it easier. If the system has no solution or infinitely many solu-
tions, do not complete the row.
Step 4 Find the maximum value of P over those completed rows that have no
negative values. It is the optimal value of the objective function, provided
the optimal value exists.
Table 1 shows the table of step 2 for the tent production problem. Note that
m = 2, so there are 1m + 221m + 12 >2 = 6 rows and m + 2 = 4 columns.
Table 1
x1 x2 s1 s2
0 0
0 0
0 0
0 0
0 0
0 0
Table 2 shows the table of step 3: Each row of Table 1 has been completed to a
solution of the e-system (3). This involves solving the e-system six different times.
Of course, the solution in the first row is easy: If x1 and x2 are both assigned the value
0 in (3), then clearly s1 = 32 and s2 = 84. The remaining rows are completed simi-
larly: After two variables are assigned the value 0, the resulting system of two equa-
tions in two variables can be solved by the Gauss–Jordan method, or by substitution,
or by elimination by addition. Such solutions of (3), in which two of the variables
have been assigned the value 0, are called basic solutions. So there are six basic
solutions in Table 2. Note that the six basic solutions correspond to the six intersection
points O, A, B, C, D, and E of Figure 1.
Table 2 Basic Solutions
x1 x2 s1 s2
0 0 32 84
0 16 0 20
0 21 -10 0
32 0 0 -12
28 0 4 0
20 6 0 0
We use step 4 of the procedure to obtain the solution of the linear programming
problem. The value of P = 50x1 + 80x2 is calculated for each row of Table 2 that
has no negative values. See Table 3. We conclude that the solution of the linear pro-
gramming problem is
Max P = $1,480 at x1 = 20, x2 = 6
Table 3 The Table Method

x1 x2 s1 s2 P = 50x1 + 80x2
0 0 32 84 0
0 16 0 20 1,280
0 21 -10 0 –
32 0 0 -12 –
28 0 4 0 1,400
20 6 0 0 1,480
We have ignored the third and fourth rows because s1 = - 10 in the third row and
s2 = - 12 in the fourth. Those basic solutions are not feasible. The four remaining basic
solutions are feasible. We call them basic feasible solutions. Note that the basic feasible
solutions correspond to the four corner points O, A, B, C of the feasible region (Fig. 1).
Because basic feasible solutions correspond to the corner points of the feasible
region, we can reformulate the fundamental theorem (Theorem 1 in Section 4.3):
THEOREM 1 Fundamental Theorem of Linear Programming: Version 2

exists, then that value must occur at one or more of the basic feasible solutions.
EXAMPLE 1 Slack Variables and Basic Solutions

(A) Use slack variables s1 and s2 to convert the following i-system (system of
inequalities) to an e-system (system of equations).
3x1 + 2x2 … 21
x1 + 5x2 … 20
x1, x2 Ú 0
(B) Find the basic solution for which x1 = 0 and s1 = 0.
(C) Find the basic solution for which s1 = 0 and s2 = 0.
SOLUTION
(A) Introduce slack variables s1 and s2 (one slack variable for each of the problem
constraint inequalities):
3x1 + 2x2 + s1 = 21
(4)
x1 + 5x2 + s2 = 20
(B) Substitute x1 = 0 and s1 = 0 into e-system (4):

3102 + 2x2 + 0 = 21
0 + 5x2 + s2 = 20
Divide the first equation by 2 to get x2 = 10.5. Substitute x2 = 10.5 in the
second equation and solve for s2:
s2 = 20 - 5110.52 = - 32.5
The basic solution is
1x1, x2, s1, s2 2 = 10, 10.5, 0, - 32.52
Note that this basic solution is not feasible because at least one of the variables
(s2 in this case) has a negative value.
(C) Substitute s1 = 0 and s2 = 0 into e-system (4):
3x1 + 2x2 + 0 = 21
x1 + 5x2 + 0 = 20
This system of two equations in two variables can be solved by Gauss–Jordan elimi-
nation or by another of our standard methods. Multiplying the second equation
by - 3 and adding the two equations gives - 13x2 = - 39, so x2 = 3. Substituting
x2 = 3 in the second equation gives x1 = 5.
The basic solution is
1x1, x2, s1, s2 2 = 15, 3, 0, 02
Note that this basic solution is feasible because none of the variables has a negative
value.
Matched Problem 1 Refer to Example 1. Find the basic solution for which
x2 = 0 and s1 = 0.
EXAMPLE 2 The Table Method Construct the table of basic solutions and use it to solve the
following linear programming problem:
Maximize P = 10x1 + 25x2
subject to 3x1 + 2x2 … 21
x1 + 5x2 … 20
x1, x2 Ú 0
SOLUTION The system of inequalities is identical to the i-system of Example 1. The

number of problem constraints is m = 2, so there will be 1m + 221m + 12 >2 = 6
rows in the table of basic solutions. Because there are two decision variables, x1
and x2, we assign two zeros to each row of the table in all possible combinations.
We also include the basic solution that was found in Example 1B (row 2 of table)
and the basic solution that was found in Example 1C (last row).
x1 x2 s1 s2
0 0
0 10.5 0 -32.5
0 0
0 0
0 0
5 3 0 0
We complete the table working one row at a time. We substitute 0’s for the two
variables indicated by the row, in the e-system
3x1 + 2x2 + s1 = 21
x1 + 5x2 + s2 = 20
The result is a system of two equations in two variables, which can be solved by
Gauss–Jordan elimination or another of our standard methods. Table 4 shows all
six basic solutions and the values of the objective function P = 10x1 + 25x2 at the
four basic feasible solutions.
x1 x2 s1 s2 P = 10x1 + 25x2
0 0 21 20 0
0 10.5 0 -32.5 –
0 4 13 0 100
7 0 0 13 70
20 0 -39 0 –
5 3 0 0 125
We conclude that
Max P = 125 at x1 = 5, x2 = 3
Matched Problem 2 Construct the table of basic solutions and use it to solve
the following linear programming problem:
4x1 + 3x2 … 36
x1, x2 Ú 0

The following linear programming problem has only one problem constraint:
x1, x2 Ú 0
Solve it by the table method, then solve it by graphing, and compare the two solutions.
EXAMPLE 3 The Table Method Construct the table of basic solutions and use it to solve the
following linear programming problem:
subject to x1 + 6x2 … 72
x1 + 3x2 … 45
2x1 + 3x2 … 72
x1, x2 Ú 0
SOLUTION There are m = 3 problem constraints, so we use slack variables s1, s2,
and s3 to convert the i-system to the e-system (5):
x1 + 6x2 + s1 = 72
x1 + 3x2 + s2 = 45 (5)
2x1 + 3x2 + s3 = 72
There will be 1m + 221m + 12 >2 = 10 rows in the table of basic solutions.

Because there are two decision variables, x1 and x2, we assign two zeros to each
row of the table in all possible combinations.
x1 x2 s1 s2 s3
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
We complete the table working one row at a time. We substitute 0’s for the two vari-
ables indicated by the row, in the e-system (5). The result is a system of three equa-
tions in three variables, which can be solved by Gauss–Jordan elimination, or by
another of our standard methods. Table 5 shows all ten basic solutions, and the val-
ues of the objective function, P = 40x1 + 50x2, at the five basic feasible solutions.
x1 x2 s1 s2 s3 P = 40x1 + 50x2
0 0 72 45 72 0
0 12 0 9 36 600
0 15 -18 0 27 –
0 24 -72 -27 0 –
72 0 0 -27 -72 –
45 0 27 0 -18 –
36 0 36 9 0 1,440
18 9 0 0 9 1,170
24 8 0 -3 0 –
27 6 9 0 0 1,380
We conclude that
Max P = 1,440 at x1 = 36, x2 = 0
Matched Problem 3 Construct the table of basic solutions and use it to solve
the following linear programming problem:
x1 + x2 … 5
2x1 + x2 … 8
x1, x2 Ú 0
Basic and Nonbasic Variables

The basic solutions associated with a linear programming problem are found by
assigning the value 0 to certain decision variables (the xi’s) and slack variables
(the si’s). Consider, for example, row 2 of Table 5. That row shows the basic solution
1x1, x2, s1, s2, s3 2 = 10, 12, 0, 9, 362. It is customary to refer to the variables that
are assigned the value 0 as nonbasic variables and to the others as basic variables.
So for the basic solution of row 2, the basic variables are x2, s2, and s3; the nonbasic
variables are x1 and s1.
Note that the classification of variables as basic or nonbasic depends on the basic
solution. Row 8 of Table 5 shows the basic solution 1x1, x2, s1, s2, s32 = 118, 9, 0, 0, 92.
For row 8, the basic variables are x1, x2, and s3; the nonbasic variables are s1 and s2.
EXAMPLE 4 Basic and Nonbasic Variables Refer to Table 5. For the basic solution
1x1, x2, s1, s2, s3 2 = 136, 0, 36, 9, 02 in row 7 of Table 5, classify the variables as
basic or nonbasic.
SOLUTION The basic variables are x1, s1, and s2. The other variables, x2 and s3, were
assigned the value 0, and therefore are nonbasic.
Matched Problem 4 Refer to Table 5. For the basic solution

1x1, x2, s1, s2, s3 2 = 127, 6, 9, 0, 02 of row 10, classify the variables as basic or
nonbasic.

Use the table method to solve the following linear programming problem, and
explain why one of the rows in the table cannot be completed to a basic solution:
subject to x1 + x2 … 2
x1 + x2 … 3
x1, x2 Ú 0
Reminder Summary
The expression 6!, read “6 facto- The examples in this section illustrate the table method when there are two decision
rial,” stands for the product of the variables. But the method can be used when there are k decision variables, where k is
natural numbers from 1 through 6. So any positive integer.
6! = 6 # 5 # 4 # 3 # 2 # 1 = 720, and The number of ways in which r objects can be chosen from a set of n objects,
6!
6C2 =
without regard to order, is denoted by nCr and given by the formula
2!16 - 22!
6#5#4#3#2#1 n!
= # # # # # nCr =
2 1 4 3 2 1 r!1n - r2!
6#5 (The formula, giving the number of combinations of n distinct objects taken r at a
= # = 15.
2 1 time, is explained and derived in Chapter 6). If there are k decision variables and m
problem constraints in a linear programming problem, then the number of rows in the
table of basic solutions is k + mCk, because this is the number of ways of selecting k of
the k + m variables to be assigned the value 0.
PROCEDURE The Table Method (k Decision Variables)

Assume that a standard maximization problem in standard form has k decision
variables x1, x2, . . ., xk, and m problem constraints.
Step 1 Use slack variables s1, s2, . . ., sm to convert the i-system to an e-system.
Step 2 Form a table with k + mCk rows and k + m columns labeled x1, x2, . . ., xk,
s1, s2, . . ., sm. In the first row, assign 0 to x1, x2, . . ., xk. Continue until the
rows contain all possible combinations of assigning k 0’s to the variables.
Step 3 Complete each row to a solution of the e-system, if possible. Because k of
the variables have the value 0, this involves solving a system of m linear
equations in m variables. Use the Gauss–Jordan method, or another method
if you find it easier. If the system has no solutions, or infinitely many solu-
tions, do not complete the row.
Step 4 Find the maximum value of P over those completed rows that have no
negative values (that is, over the basic feasible solutions). It is the optimal
value of the objective function, provided the optimal value exists.
The benefit of the table method is that it gives a procedure for finding all corner
points of the feasible region without drawing a graph.
Unfortunately, the number of rows in the table becomes too large to be practical,
even for computers, when the number of decision variables and problem constraints
is large. For example, with k = 30 decision variables and m = 35 problem con-
straints, the number of rows is
65C30 ≈ 3 * 1018
We need a procedure for finding the optimal solution of a linear programming
problem without having to find every corner point. The simplex method, discussed
in the next section, is such a procedure. It gives a practical method for solving large
linear programming problems.
Exercises 5.1
Skills Warm-up Exercises 9. Find the solution of the system for which x1 = 0, s1 = 0.
W In Problems 1–8, evaluate the expression. (If necessary, review 10. Find the solution of the system for which x1 = 0, s2 = 0.
Section C.3). 11. Find the solution of the system for which x2 = 0, s2 = 0.
8! 10! 11! 7! 12. Find the solution of the system for which x2 = 0, s1 = 0.
1. 2. 3. 4.
3!5! 2!8! 9!2! 4!3!
In Problems 13–20, write the e-system obtained via slack
5. In how many ways can two variables be chosen from x1, x2, variables for the given linear programming problem.
s1, s2, s3, s4, s5 and assigned the value 0? 13. Maximize P = 5x1 + 7x2
6. In how many ways can three variables be chosen from x1, x2, subject to 2x1 + 3x2 … 9
x3, s1, s2, s3 and assigned the value 0? 6x1 + 7x2 … 13
7. In how many ways can four variables be chosen from x1, x2, x1, x2 Ú 0
x3, x4, s1, s2, s3, s4 and assigned the value 0?
14. Maximize P = 3x1 + 9x2
8. In how many ways can three variables be chosen from x1, x2, subject to x1 + 3x2 … 60
s1, s2, s3, s4, s5, s6, and assigned the value 0?
x1 - x2 … 0
A Problems 9–12 refer to the system x1, x2 Ú 0
2x1 + 5x2 + s1 = 10
x1 + 3x2 + s2 = 8
15. Maximize P = 3x1 + 5x2 27. Use the basic feasible solutions to maximize P = 2x1 + 5x2.
subject to 12x1 - 14x2 … 55 28. Use the basic feasible solutions to maximize P = 8x1 + 5x2.
19x1 + 5x2 … 40
29. Describe geometrically the set of all points in the plane such
- 8x1 + 11x2 … 64 that s1 7 0.
x1, x2 Ú 0 30. Describe geometrically the set of all points in the plane such
16. Maximize P = 2x1 + x2 that s2 6 0.
subject to x1 - 3x2 … 3
Problems 31–40 refer to the partially completed table below of the
3x1 + 4x2 … 24 10 basic solutions to the e-system
- 3x1 + 2x2 … 6 x1 + x2 + s1 = 24
x1, x2 Ú 0
2x1 + x2 + s2 = 30
4x1 + x2 + s3 = 48
x1, x2 Ú 0
x1 x2 s1 s2 s3
18. Maximize P = 13x1 + 8x2 (A) 0 0 24 30 48
subject to x1 + 2x2 … 20 (B) 0 24 0 6 24
x1, x2 Ú 0 (C) 0 30 -6 0 18
19. Maximize P = x1 + 2x2 (D) 0 48 -24 -18 0
subject to 4x1 - 3x2 … 12 (E) 24 0 0 -18 -48
5x1 + 2x2 … 25 (F) 15 0 9 0 -12

(G) 0 0
- 3x1 + 7x2 … 32
(H) 0 0
2x1 + x2 … 9
(I) 0 0
x1, x2 Ú 0
(J) 0 0
subject to 30x1 - 25x2 … 75 31. In basic solution (C), which variables are basic?
10x1 + 13x2 … 30 32. In basic solution (E), which variables are nonbasic?
5x1 + 18x2 … 40
33. In basic solution (G), which variables are nonbasic?
40x1 + 36x2 … 85
34. In basic solution (I), which variables are basic?
x1, x2 Ú 0
35. Which of the basic solutions (A) through (F) are not feasible?
B Problems 21–30 refer to the table below of the six basic solutions Explain.
to the e-system
36. Which of the basic solutions (A) through (F) are feasible?
2x1 + 3x2 + s1 = 24 Explain.
4x1 + 3x2 + s2 = 36 37. Find basic solution (G).
x1 x2 s1 s2 38. Find basic solution (H).
(A) 0 0 24 36
39. Find basic solution (I).
(B) 0 8 0 12
40. Find basic solution (J).
(C) 0 12 -12 0
(D) 12 0 0 -12 In Problems 41–48, convert the given i-system to an e-system
(E) 9 0 6 0 using slack variables. Then construct a table of all basic solutions
(F) 6 4 0 0 of the e-system. For each basic solution, indicate whether or not it
is feasible.
21. In basic solution (A), which variables are basic? 41. 4x1 + 5x2 … 20 42. 6x1 + 5x2 … 30
22. In basic solution (B), which variables are nonbasic? x1, x2 Ú 0 x1, x2 Ú 0
23. In basic solution (C), which variables are nonbasic? 43. x1 + x2 … 6 44. 5x1 + x2 … 15
24. In basic solution (D), which variables are basic? x1 + 4x2 … 12 x1 + x2 … 7
25. Which of the six basic solutions are feasible? Explain. x1, x2 Ú 0 x1, x2 Ú 0
26. Which of the basic solutions are not feasible? Explain.
45. 2x1 + 5x2 … 20 46. 5x1 + 6x2 … 45 63. Maximize P = 25x1 + 10x2
x1 + 2x2 … 9 3x1 + 2x2 … 18 subject to 2x1 + 5x2 … 20
x1 + 2x2 … 9
x1, x2 Ú 0 x1, x2 Ú 0
x1, x2 Ú 0
47. x1 + 2x2 … 24 48. 5x1 + 4x2 … 240
64. Maximize P = 100x1 + 40x2
x1 + x2 … 15 5x1 + 2x2 … 150
2x1 + x2 … 24 5x1 + x2 … 120 3x1 + 2x2 … 18
x1, x2 Ú 0 x1, x2 Ú 0 x1, x2 Ú 0
In Problems 49–54, graph the system of inequalities from the 65. Maximize P = 30x1 + 40x2
given problem, and list the corner points of the feasible region. subject to x1 + 2x2 … 24
Verify that the corner points of the feasible region correspond to x1 + x2 … 15
the basic feasible solutions of the associated e-system.
2x1 + x2 … 24
49. Problem 41 50. Problem 42 x1, x2 Ú 0
51. Problem 43 52. Problem 44 66. Maximize P = x1 + x2
53. Problem 45 54. Problem 46 subject to 5x1 + 4x2 … 240
C 55. For a standard maximization problem in standard form, with 5x1 + 2x2 … 150
two decision variables x1 and x2, explain why the feasible 5x1 + x2 … 120
region is not empty. x1, x2 Ú 0
56. For a standard maximization problem in standard form, with
k decision variables, x1, x2, . . ., xk, explain why the feasible In Problems 67–70, explain why the linear programming problem
region is not empty. has no optimal solution.
57. If 5x1 + 4x2 … 1,000 is one of the problem constraints in a 67. Maximize P = 8x1 + 9x2
standard maximization problem in standard form with two subject to 3x1 - 7x2 … 42
decision variables, explain why the optimal value of the ob- x1, x2 Ú 0
jective function exists. [Hint: See Theorem 2 in Section 4.3].
58. If a1x1 + a2x2 … b is one of the problem constraints in a subject to - 2x1 + 10x2 … 30
standard maximization problem in standard form with two
decision variables, and a1 and a2 are both positive, explain x1, x2 Ú 0
why the optimal value of the objective function exists. [Hint: 69. Maximize P = 6x1 + 13x2
See Theorem 2 in Section 4.3]. subject to - 4x1 + x2 … 4
In Problems 59–66, solve the given linear programming problem 4x1 - 5x2 … 12
using the table method (the table of basic solutions was constructed x1, x2 Ú 0
in Problems 41–48).
70. Maximize P = 18x1 + 11x2
59. Maximize P = 10x1 + 9x2 subject to 3x1 - 2x2 … 6
subject to 4x1 + 5x2 … 20 - 3x1 + 2x2 … 6
x1, x2 Ú 0 x1, x2 Ú 0
60. Maximize P = 3x1 + 8x2 In Problems 71–72, explain why the linear programming problem
subject to 6x1 + 5x2 … 30 has an optimal solution, and find it using the table method.
x1, x2 Ú 0 71. Maximize P = 8x1 + 10x2
61. Maximize P = 15x1 + 20x2 subject to x1 + x2 … 7000
subject to x1 + x2 … 6 x2 … 3000
x1 + 4x2 … 12 x1, x2 Ú 0
x1, x2 Ú 0 72. Maximize P = 5x1 + 3x2
62. Maximize P = 5x1 + 20x2 subject to x1 + x2 … 300
subject to 5x1 + x2 … 15 2x1 + x2 … 360
x1 + x2 … 7 x1, x2 Ú 0
x1, x2 Ú 0 73. A linear programming problem has four decision variables
x1, x2, x3, x4, and six problem constraints. How many rows are
SECTION 5.2 The Simplex Method: Maximization with Problem Constraints of the Form … 277
there in the table of basic solutions of the associated 3. x1 x2 s1 s2 s3 P = 36x1 + 24x2

e-system? 0 0 8 5 8 0
74. A linear programming problem has five decision variables x1, 0 4 0 1 4 96
x2, x3, x4, x5 and six problem constraints. How many rows are 0 5 -2 0 3 –
there in the table of basic solutions of the associated e-system?
0 8 -8 -3 0 –
75. A linear programming problem has 30 decision variables 8 0 0 -3 -8 –
x1, x2, c , x30 and 42 problem constraints. How many
5 0 3 0 -2 –
rows are there in the table of basic solutions of the associated
e-system? (Write the answer using scientific notation.) 4 0 4 1 0 144
2 3 0 0 1 144
76. A linear programming problem has 40 decision variables
8>3 8>3 0 -1>3 0 –
x1, x2, c , x40 and 85 problem constraints. How many rows
are there in the table of basic solutions of the associated 3 2 1 0 0 156
e-system? (Write the answer using scientific notation.) Max P = 156 at x1 = 3, x2 = 2
4. x1, x2, and s1 are basic; s2 and s3 are nonbasic
1. 1x1, x2, s1, s2 2 = 17, 0, 0, 132

2. x1 x2 s1 s2 P = 30x + 40x
0 0 24 36 0
0 8 0 12 320
0 12 -12 0 –
12 0 0 -12 –
9 0 6 0 270
6 4 0 0 340
Max P = 340 at x1 = 6, x2 = 4
5.2 The Simplex Method:

Maximization with Problem Constraints of the Form …
■■ Initial System Now we can develop the simplex method for a standard maximization problem. The
■■ Simplex Tableau
simplex method is most useful when used with computers. Consequently, it is not
intended that you become an expert in manually solving linear programming prob-
■■ Pivot Operation lems using the simplex method. But it is important that you become proficient in
■■ Interpreting the Simplex Process constructing the models for linear programming problems so that they can be solved
Geometrically using a computer, and it is also important that you develop skill in interpreting the
■■ Simplex Method Summarized
results. One way to gain this proficiency and interpretive skill is to set up and manu-
ally solve a number of fairly simple linear programming problems using the simplex
■■ Application method. This is the main goal in this section and in Sections 5.3 and 5.4. To assist
you in learning to develop the models, the answer sections for Exercises 5.2, 5.3, and
5.4 contain both the model and its solution.
Initial System
We will introduce the concepts and procedures involved in the simplex method
through an example—the tent production example discussed earlier. We restate the
problem here in standard form for convenient reference:
Maximize P = 50x1 + 80x2 Objective function
f Problem constraints (1)
3x1 + 4x2 … 84
Introducing slack variables s1 and s2, we convert the problem constraint inequalities
in problem (1) into the following system of problem constraint equations:
x1 + 2x2 + s1 = 32
3x1 + 4x2 + s2 = 84 (2)
x1, x2, s1, s2 Ú 0
Since a basic solution of system (2) is not feasible if it contains any negative values,
we have also included the nonnegative constraints for both the decision variables x1
and x2 and the slack variables s1 and s2. From our discussion in Section 5.1, we know
that out of the infinitely many solutions to system (2), an optimal solution is one of the
basic feasible solutions, which correspond to the corner points of the feasible region.
As part of the simplex method we add the objective function equation
P = 50x1 + 80x2 in the form - 50x1 - 80x2 + P = 0 to system (2) to create what
is called the initial system:
x1 + 2x2 + s1 = 32
3x1 + 4x2 + s2 = 84
(3)
- 50x1 - 80x2 + P = 0
x1, x2, s1, s2 Ú 0
When we add the objective function equation to system (2), we must slightly
modify the earlier definitions of basic solution and basic feasible solution so that they
apply to the initial system (3).
DEFINITION Basic Solutions and Basic Feasible Solutions for Initial Systems
1. The objective function variable P is always selected as a basic variable.
2. Note that a basic solution of system (3) is also a basic solution of system (2)
after P is deleted.
3. If a basic solution of system (3) is a basic feasible solution of system (2) after
deleting P, then the basic solution of system (3) is called a basic feasible
solution of system (3).
4. A basic feasible solution of system (3) can contain a negative number, but only
if it is the value of P, the objective function variable.
These changes lead to a small change in the second version of the fundamental
theorem (see Theorem 1, Section 5.1).
THEOREM 1 Fundamental Theorem of Linear Programming: Version 3

exists, then that value must occur at one or more of the basic feasible solutions of
the initial system.
With these adjustments understood, we start the simplex process with a basic
feasible solution of the initial system (3), which we will refer to as an initial basic
feasible solution. An initial basic feasible solution that is easy to find is the one
associated with the origin.
Since system (3) has three equations and five variables, it has three basic vari-
ables and two nonbasic variables. Looking at the system, we see that x1 and x2 appear
in all equations, and s1, s2, and P each appear only once and each in a different equa-
tion. A basic solution can be found by inspection by selecting s1, s2, and P as the
basic variables (remember, P is always selected as a basic variable) and x1 and x2 as
the nonbasic variables to be set equal to 0. Setting x1 and x2 equal to 0 and solving for
the basic variables, we obtain the basic solution:
x1 = 0, x2 = 0, s1 = 32, s2 = 84, P = 0
This basic solution is feasible since none of the variables (excluding P) are negative.
This is the initial basic feasible solution that we seek.
Now you can see why we wanted to add the objective function equation to
system (2): A basic feasible solution of system (3) not only includes a basic feasible
solution of system (2), but, in addition, it includes the value of P for that basic fea-
sible solution of system (2).
The initial basic feasible solution we just found is associated with the origin. Of
course, if we do not produce any tents, we do not expect a profit, so P = $0. Starting
with this easily obtained initial basic feasible solution, the simplex process moves through
each iteration (repetition) to another basic feasible solution, each time improving the
profit. The process continues until the maximum profit is reached, then the process stops.
Simplex Tableau
To facilitate the search for the optimal solution, we turn to matrix methods discussed
in Chapter 3. Our first step is to write the augmented matrix for the initial system (3).
This matrix is called the initial simplex tableau, and it is simply a tabulation of the
coefficients in system (3).
x1 x2 s1 s2 P
s1 1 2 1 0 0 32
s2 C 3 4 0 1 0 3 84 S Initial simplex tableau (4)
P - 50 - 80 0 0 1 0
In tableau (4), the row below the dashed line always corresponds to the objective
function. Each of the basic variables we selected above, s1, s2, and P, is also placed on
the left of the tableau so that the intersection element in its row and column is not 0.
For example, we place the basic variable s1 on the left so that the intersection ele-
ment of the s1 row and the s1 column is 1 and not 0. The basic variable s2 is similarly
placed. The objective function variable P is always placed at the bottom. The reason
for writing the basic variables on the left in this way is that this placement makes it
possible to read certain basic feasible solutions directly from the tableau. If x1 = 0
and x2 = 0, the basic variables on the left of tableau (4) are lined up with their cor-
responding values, 32, 84, and 0, to the right of the vertical line.
Looking at tableau (4) relative to the choice of s1, s2, and P as basic variables,
we see that each basic variable is above a column that has all 0 elements except for
a single 1 and that no two such columns contain 1’s in the same row. These observa-
tions lead to a formalization of the process of selecting basic and nonbasic variables
that is an important part of the simplex method:
PROCEDURE Selecting Basic and Nonbasic Variables for the Simplex

Method
Given a simplex tableau,
Step 1 Numbers of variables. Determine the number of basic variables and the
number of nonbasic variables. These numbers do not change during the
simplex process.
Step 2 Selecting basic variables. A variable can be selected as a basic variable only
if it corresponds to a column in the tableau that has exactly one nonzero ele-
ment (usually 1) and the nonzero element in the column is not in the same
row as the nonzero element in the column of another basic variable. This
procedure always selects P as a basic variable, since the P column never

changes during the simplex process.
Step 3 Selecting nonbasic variables. After the basic variables are selected in
step 2, the remaining variables are selected as the nonbasic variables. The
tableau columns under the nonbasic variables usually contain more than
one nonzero element.
The earlier selection of s1, s2, and P as basic variables and x1 and x2 as nonbasic
variables conforms to this prescribed convention of selecting basic and nonbasic vari-
ables for the simplex process.
Pivot Operation
The simplex method swaps one of the nonbasic variables, x1 or x2, for one of the
basic variables, s1 or s2 (but not P), as a step toward improving the profit. For a non-
basic variable to be classified as a basic variable, we need to perform appropriate row
operations on the tableau so that the newly selected basic variable will end up with
exactly one nonzero element in its column. In this process, the old basic variable will
usually gain additional nonzero elements in its column as it becomes nonbasic.
Which nonbasic variable should we select to become basic? It makes sense to
select the nonbasic variable that will increase the profit the most per unit change in
that variable. Looking at the objective function
P = 50x1 + 80x2
we see that if x1 stays a nonbasic variable (set equal to 0) and if x2 becomes a new
basic variable, then
P = 50102 + 80x2 = 80x2
and for each unit increase in x2, P will increase $80. If x2 stays a nonbasic variable
and x1 becomes a new basic variable, then (reasoning in the same way) for each unit
increase in x1, P will increase only $50. So we select the nonbasic variable x2 to
enter the set of basic variables, and call it the entering variable. (The basic variable
leaving the set of basic variables to become a nonbasic variable is called the exiting
variable, which will be discussed shortly.)
The column corresponding to the entering variable is called the pivot column.
Looking at the bottom row in tableau (4)—the objective function row below the dashed
line—we see that the pivot column is associated with the column to the left of the P
column that has the most negative bottom element. In general, the most negative element
in the bottom row to the left of the P column indicates the variable above it that will pro-
duce the greatest increase in P for a unit increase in that variable. For this reason, we call
the elements in the bottom row of the tableau, to the left of the P column, indicators.
We illustrate the indicators, the pivot column, the entering variable, and the ini-
tial basic feasible solution below:
Entering
variable
T
x1 x2 s1 s2 P
s1 1 2 1 0 0 32
s2 C 3 4 0 1 0 3 84 S Initial simplex tableau
(5)
P - 50 - 80 0 0 1 0
c Indicators are shown in color.
Pivot
column
x1 = 0, x2 = 0, s1 = 32, s2 = 84, P = 0 Initial basic feasible solution
Now that we have chosen the nonbasic variable x2 as the entering variable
(the nonbasic variable to become basic), which of the two basic variables, s1 or s2,
should we choose as the exiting variable (the basic variable to become nonbasic)?
We saw above that for x1 = 0, each unit increase in the entering variable x2 results
in an increase of $80 for P. Can we increase x2 without limit? No! A limit is im-
posed by the nonnegative requirements for s1 and s2. (Remember that if any of the
basic variables except P become negative, we no longer have a feasible solution.)
So we rephrase the question and ask: How much can x2 be increased when x1 = 0
without causing s1 or s2 to become negative? To see how much x2 can be increased,
we refer to tableau (5) or system (3) and write the two problem constraint equa-
tions with x1 = 0:
2x2 + s1 = 32
4x2 + s2 = 84
Solving for s1 and s2, we have
s1 = 32 - 2x2
s2 = 84 - 4x2
For s1 and s2 to be nonnegative, x2 must be chosen so that both 32 - 2x2 and 84 - 4x2
are nonnegative. That is, so that
32 - 2x2 Ú 0 and 84 - 4x2 Ú 0
- 2x2 Ú - 32 - 4x2 Ú - 84
x2 … 32
2 = 16 x2 … 84
4 = 21
For both inequalities to be satisfied, x2 must be less than or equal to the smaller of
the values, which is 16. So x2 can increase to 16 without either s1 or s2 becoming
negative. Now, observe how each value (16 and 21) can be obtained directly from the
following tableau:
Entering
variable
T
x1 x2 s1 s2 P
32
s1 1 2 1 0 0 32 2 = 16 (smallest)
s2 C 3 4 0 1 0 3 84 S 84
4 = 21 (6)
P - 50 - 80 0 0 1 0
c
Pivot
column
From tableau (6) we can determine the amount that the entering variable
can increase by choosing the smallest of the quotients obtained by dividing each
element in the last column above the dashed line by the corresponding positive
element in the pivot column. The row with the smallest quotient is called the pivot
row, and the variable to the left of the pivot row is the exiting variable. In this
case, s1 will be the exiting variable, and the roles of x2 and s1 will be interchanged.
The element at the intersection of the pivot column and the pivot row is called the
pivot element, and we circle this element for ease of recognition. Since a negative
or 0 element in the pivot column places no restriction on the amount that an enter-
ing variable can increase, it is not necessary to compute the quotient for negative
or 0 values in the pivot column.
A negative or 0 element is never selected for the pivot element.
The following tableau illustrates this process, which is summarized in the next
box.
Entering
variable
Pivot element T
x1 n x2 s1 s2 P
Exiting variable S s1 1 2 1 0 0 32 d Pivot row
s2 C 3 4 0 1 0 3 84 S (7)
P - 50 - 80 0 0 1 0
c
Pivot
column
PROCEDURE Selecting the Pivot Element

Step 1 Locate the most negative indicator in the bottom row of the tableau to the
left of the P column (the negative number with the largest absolute value).
The column containing this element is the pivot column. If there is a tie for
the most negative indicator, choose either column.
Step 2 Divide each positive element in the pivot column above the dashed line into
the corresponding element in the last column. The pivot row is the row cor-
responding to the smallest quotient obtained. If there is a tie for the smallest
quotient, choose either row. If the pivot column above the dashed line has no
positive elements, there is no solution, and we stop.
Step 3 The pivot (or pivot element) is the element at the intersection of the pivot
column and pivot row.
Note: The pivot element is always positive and never appears in the bottom row.
Remember: The entering variable is at the top of the pivot column, and the exiting
variable is at the left of the pivot row.
In order for x2 to be classified as a basic variable, we perform row operations on

tableau (7) so that the pivot element is transformed into 1 and all other elements in the
column into 0’s. This procedure for transforming a nonbasic variable into a basic vari-
able is called a pivot operation, or pivoting, and is summarized in the following box.
PROCEDURE Performing a Pivot Operation

A pivot operation, or pivoting, consists of performing row operations as follows:
Step 1 Multiply the pivot row by the reciprocal of the pivot element to transform
the pivot element into a 1. (If the pivot element is already a 1, omit this step.)
Step 2 Add multiples of the pivot row to other rows in the tableau to transform all
other nonzero elements in the pivot column into 0’s.
CONCEPTUAL INSIGHT
A pivot operation uses some of the same row operations as those used in Gauss–
Jordan elimination, but there is one essential difference. In a pivot operation,
you can never interchange two rows.
Performing a pivot operation has the following effects:

1. The (entering) nonbasic variable becomes a basic variable.
2. The (exiting) basic variable becomes a nonbasic variable.
3. The value of the objective function is increased, or, in some cases, remains the same.
We now carry out the pivot operation on tableau (7). (To facilitate the process,
we do not repeat the variables after the first tableau, and we use “Enter” and “Exit”
for “Entering variable” and “Exiting variable,” respectively.)
Enter
T
x1 x2 s1 s2 P
1
Exit S s1 1 2 1 0 0 32 2 Rt S R1
s2 C 3 4 0 1 0 3 84 S
P - 50 - 80 0 0 1 0
1 1
2 1 2 0 0 16
∼C 3 4 0 1 0 † 84 S 1 - 42R1 + R2 S R2
- 50 - 80 0 0 1 0 80R1 + R3 S R3
1 1
2 1 2 0 0 16
∼C 1 0 -2 1 0 † 20 S
- 10 0 40 0 1 1,280
We have completed the pivot operation, and now we must insert appropriate
variables for this new tableau. Since x2 replaced s1, the basic variables are now x2, s2,
and P, as indicated by the labels on the left side of the new tableau. Note that this
selection of basic variables agrees with the procedure outlined on pages 279 and 280
for selecting basic variables. We write the new basic feasible solution by setting the
nonbasic variables x1 and s1 equal to 0 and solving for the basic variables by inspec-
tion. (Remember, the values of the basic variables listed on the left are the corre-
sponding numbers to the right of the vertical line. To see this, substitute x1 = 0 and
s1 = 0 in the corresponding system shown next to the simplex tableau.)
x1 x2 s1 s2 P
1 1 1 1
x2 2 1 2 0 0 16 2 x1 + x2 + 2 s1 = 16
s2 C 1 0 -2 1 0 † 20 S x1 - 2s1 + s2 = 20
P - 10 0 40 0 1 1,280 - 10x 1 + 40s 1 + P = 1,280
x1 = 0, x2 = 16, s1 = 0, s2 = 20, P = $1,280
A profit of $1,280 is a marked improvement over the $0 profit produced by the

initial basic feasible solution. But we can improve P still further, since a negative indi-
cator still remains in the bottom row. To see why, we write out the objective function:
- 10x1 + 40s1 + P = 1,280
or
P = 10x1 - 40s1 + 1,280
If s1 stays a nonbasic variable (set equal to 0) and x1 becomes a new basic vari-
able, then
P = 10x1 - 40102 + 1,280 = 10x1 + 1,280
and for each unit increase in x1, P will increase $10.
We now go through another iteration of the simplex process using another pivot
element. The pivot element and the entering and exiting variables are shown in the
following tableau:
Enter
T
x1 x2 s1 s2 P
1 1 16
x2 2 1 2 0 0 16 1 > 2 = 32
Exit S s2 C 1 0 -2 1 0 † 20 S 20
1 = 20
P - 10 0 40 0 1 1,280
We now pivot on (the circled) 1. That is, we perform a pivot operation using this
1 as the pivot element. Since the pivot element is 1, we do not need to perform the
first step in the pivot operation, so we proceed to the second step to get 0’s above and
below the pivot element 1. As before, to facilitate the process, we omit writing the
variables, except for the first tableau.
Enter
T
x1 x2 s1 s2 P
1 1
x2 2 1 2 0 0 16 1- 12 2R2 + R1 S R1
Exit S s2 C 1 0 -2 1 0 † 20 S
P - 10 0 40 0 1 1,280 10R2 + R3 S R3
3
0 1 2 - 12 0 6
∼C 1 0 -2 1 0 † 20 S
0 0 20 10 1 1,480
Since there are no more negative indicators in the bottom row, we are done. Let
us insert the appropriate variables for this last tableau and write the corresponding
basic feasible solution. The basic variables are now x1, x2, and P, so to get the cor-
responding basic feasible solution, we set the nonbasic variables s1 and s2 equal to 0
and solve for the basic variables by inspection.
x1 x2 s1 s2 P
3
x2 0 1 - 12 02 6
x1 C 1 0 -2 1 0 † 20 S
P 0 0 20 10 1 1,480
x1 = 20, x2 = 6, s1 = 0, s2 = 0, P = 1,480
To see why this is the maximum, we rewrite the objective function from the
bottom row:
20s1 + 10s2 + P = 1,480
P = 1,480 - 20s1 - 10s2
Since s1 and s2 cannot be negative, any increase of either variable from 0 will make
the profit smaller.
Finally, returning to our original problem, we conclude that a production schedule
of 20 standard tents and 6 expedition tents will produce a maximum profit of $1,480 per
day, just as was found by the geometric method in Section 4.3. The fact that the slack
variables are both 0 means that, for this production schedule, the plant will operate at full
capacity—there is no slack in either the cutting department or the assembly department.
Interpreting the Simplex Process Geometrically

We can interpret the simplex process geometrically in terms of the feasible region
graphed in the preceding section. Table 1 lists the three basic feasible solutions we
just found using the simplex method (in the order they were found). Table 1 also in-
cludes the corresponding corner points of the feasible region illustrated in Figure 1.
Table 1 Basic Feasible Solution (obtained above) x2
x1 x2 s1 s2 P1$2 Corner Point

20
0 0 32 84 0 O10, 02
A(0, 16)
0 16 0 20 1,280 A10, 162
20 6 0 0 1,480 B120, 62 10
B(20, 6)
Feasible region
C(28, 0)
x1
O(0, 0) 10 20 30 40
Figure 1
Looking at Table 1 and Figure 1, we see that the simplex process started at the
origin, moved to the adjacent corner point A10, 162, and then to the optimal solution
B120, 62 at the next adjacent corner point. This is typical of the simplex process.
Simplex Method Summarized

Before presenting additional examples, we summarize the important parts of the sim-
plex method schematically in Figure 2.
Step 1:
Write the standard maximization
problem in standard form, introduce
slack variables and the objective
function equation to form the initial
system, and write the initial tableau.
Step 2:
Are there any negative No STOP
indicators in the The optimal solution has
bottom row? been found.
Yes
Step 3:
Select the pivot column.
Step 4:
Are there any positive STOP
No
elements in the pivot The linear programming
Figure 2 Simplex algorithm for column above the problem has no optimal solution.
dashed line?
standard maximization problems
(Problem constraints are of the "
Yes
form with nonnegative constants
on the right. The coefficients of the Step 5:
objective function can be any real Select the pivot element and
perform the pivot operation.
numbers.)
EXAMPLE 1 Using the Simplex Method Solve the following linear programming problem us-
ing the simplex method:

subject to 4x1 + x2 … 28
2x1 + 3x2 … 24
x1, x2 Ú 0
SOLUTION Introduce slack variables s1 and s2, and write the initial system:
4x1 + x2 + s1 = 28
2x1 + 3x2 + s2 = 24
- 10x1 - 5x2 + P = 0
x1, x2, s1, s2 Ú 0
Write the simplex tableau, and identify the first pivot element and the entering and
exiting variables:
Enter
T
x1 x2 s1 s2 P
28
Exit S s1 4 1 1 0 0 28 4 = 7
24
s2 C 2 3 0 1 0 † 24 S 2 = 12
P - 10 -5 0 0 1 0
Perform the pivot operation:
Enter
T
x1 x2 s1 s2 P
Exit S s1 4 1 1 0 0 28 1
4 R1 S R1
s2 C 2 3 0 1 0 † 24 S
P - 10 -5 0 0 1 0
1 0.25 0.25 0 0 7
∼C 2 3 0 1 0 † 24 S 1 - 22R1 + R2 S R2
- 10 -5 0 0 1 0 10R1 + R3 S R3
x1 1 0.25 0.25 0 0 7
s
∼ 2 C 0 2.5 - 0.5 1 0 † 10 S
P 0 - 2.5 2.5 0 1 70
Since there is still a negative indicator in the last row, we repeat the process by
finding a new pivot element:
Enter
T
x1 x2 s1 s2 P
7
x1 1 0.25 0.25 0 0 7 0.25 = 28
Exit S s2 C 0 2.5 - 0.5 1 0 † 10 S 10
2.5 = 4
P 0 - 2.5 2.5 0 1 70
Performing the pivot operation, we obtain

Enter
T
x1 x2 s1 s2 P
x1 1 0.25 0.25 0 0 7
1
2.5 R2 S R2
Exit S s2 C 0 2.5 - 0.5 1 0 † 10 S
P 0 - 2.5 2.5 0 1 70
1 0.25 0.25 0 0 7 1 - 0.252R2 + R1 S R1

C
∼ 0 1 - 0.2 0.4 0 † 4 S
0 - 2.5 2.5 0 1 70 2.5R2 + R3 S R3
x1 1 0 0.3 - 0.1 0 6
∼ x2 C 0 1 - 0.2 0.4 0 † 4 S
P 0 0 2 1 1 80
Since all the indicators in the last row are nonnegative, we stop and read the optimal
solution:
Max P = 80 at x1 = 6, x2 = 4, s1 = 0, s2 = 0
(To see why this makes sense, write the objective function corresponding to the last
row to see what happens to P when you try to increase s1 or s2.)
Matched Problem 1 Solve the following linear programming problem using

the simplex method:
Maximize P = 2x1 + x2
x1 + x2 … 5
x1, x2 … 0

Graph the feasible region for the linear programming problem in Example 1 and trace
the path to the optimal solution determined by the simplex method.
EXAMPLE 2 Using the Simplex Method Solve using the simplex method:

subject to - 2x1 + 3x2 … 9
- x1 + 3x2 … 12
x1, x2 Ú 0
SOLUTION Write the initial system using the slack variables s1 and s2:
- 2x1 + 3x2 + s1 = 9
- x1 + 3x2 + s2 = 12
- 6x1 - 3x2 + P = 0
Write the simplex tableau and identify the first pivot element:
x1 x2 s1 s2 P
s1 - 2 3 1 0 0 9
s2 C - 1 3 0 1 0 3 12 S
P -6 -3 0 0 1 0
c
Pivot column
Since both elements in the pivot column above the dashed line are negative, we are
unable to select a pivot row. We stop and conclude that there is no optimal solution.
Matched Problem 2 Solve using the simplex method:

subject to - 3x1 + 4x2 … 12
x2 … 2
x1, x2 Ú 0
Refer to Examples 1 and 2. In Example 1 we concluded that we had found the

optimal solution because we could not select a pivot column. In Example 2 we con-
cluded that the problem had no optimal solution because we selected a pivot column
and then could not select a pivot row. Notice that we do not try to continue with the
simplex method by selecting a negative pivot element or using a different column for
the pivot column. Remember:
If it is not possible to select a pivot column, then the simplex method stops
and we conclude that the optimal solution has been found. If the pivot col-
umn has been selected and it is not possible to select a pivot row, then the
simplex method stops and we conclude that there is no optimal solution.
Application
EXAMPLE 3 Agriculture A farmer owns a 100-acre farm and plans to plant at most three crops.
The seed for crops A, B, and C costs $40, $20, and $30 per acre, respectively. A maxi-
mum of $3,200 can be spent on seed. Crops A, B, and C require one, two, and one work
days per acre, respectively, and there are a maximum of 160 work days available. If the
farmer can make a profit of $100 per acre on crop A, $300 per acre on crop B, and $200
per acre on crop C, how many acres of each crop should be planted to maximize profit?
SOLUTION The farmer must decide on the number of acres of each crop that should
be planted. So the decision variables are
x1 = number of acres of crop A
x2 = number of acres of crop B
x3 = number of acres of crop C
The farmer’s objective is to maximize profit:
P = 100x1 + 300x2 + 200x3
The farmer is limited by the number of acres available for planting, the money avail-
able for seed, and the available work days. These limitations lead to the following
constraints:
x1 + x2 + x3 … 100 Acreage constraint
40x1 + 20x2 + 30x3 … 3,200 Monetary constraint
x1 + 2x2 + x3 … 160 Labor constraint
Adding the nonnegative constraints, we have the following model for a linear pro-
gramming problem:
Maximize P = 100x1 + 300x2 + 200x3 Objective function
subject to x1 + x2 + x3 … 100
40x1 + 20x2 + 30x3 … 3,200 s Problem constraints
x1 + 2x2 + x3 … 160
x1, x2, x3 Ú 0 Nonnegative constraints
Next, we introduce slack variables and form the initial system:

x1 + x2 + x3 + s1 = 100
40x1 + 20x2 + 30x3 + s2 = 3,200
x1 + 2x2 + x3 + s3 = 160
- 100x1 - 300x2 - 200x3 + P = 0
x1, x2, x3, s1, s2, s3 Ú 0
Notice that the initial system has 7 - 4 = 3 nonbasic variables and 4 basic vari-
ables. Now we form the simplex tableau and solve by the simplex method:
Enter
T
x1 x2 x3 s1 s2 s3 P
s1 1 1 1 1 0 0 0 100
s2 40 20 30 0 1 0 0 3,200
≥ ∞ ¥
Exit S s3 1 2 1 0 0 1 0 160 0.5R3 S R3
P - 100 - 300 - 200 0 0 0 1 0
1 1 1 1 0 0 0 100 1 - 12R3 + R1 S R1
40 20 30 0 1 0 0 3,200 1 - 202R3 + R2 S R2
∼≥ ∞ ¥
0.5 1 0.5 0 0 0.5 0 80
- 100 - 300 - 200 0 0 0 1 0 300R3 + R4 S R4
Enter
T
x1 x2 x3 s1 s2 s3 P
Exit S s1 0.5 0 0.5 1 0 - 0.5 0 20 2R1 S R1
s2 30 0 20 0 1 - 10 0 1,600
≥ ∞ ¥
x2 0.5 1 0.5 0 0 0.5 0 80
P 50 0 - 50 0 0 150 1 24,000
1 0 1 2 0 -1 0 40
30 0 20 0 1 - 10 0 1,600 1 - 202R1 + R2 S R2
∼≥ ∞ ¥
0.5 1 0.5 0 0 0.5 0 80 1 - 0.52R1 + R3 S R3
50 0 - 50 0 0 150 1 24,000 50R1 + R4 S R4
x3 1 0 1 2 0 -1 0 40
s2 10 0 0 - 40 1 10 0 800
∼ ≥ ∞ ¥
x2 0 1 0 -1 0 1 0 60
P 100 0 0 100 0 100 1 26,000
All indicators in the bottom row are nonnegative, and now we can read the optimal
solution:
x1 = 0, x2 = 60, x3 = 40, s1 = 0, s2 = 800, s3 = 0, P = $26,000
So if the farmer plants 60 acres in crop B, 40 acres in crop C, and no crop A, the
maximum profit of $26,000 will be realized. The fact that s2 = 800 tells us (look at
the second row in the equations at the start) that this maximum profit is reached by
using only $2,400 of the $3,200 available for seed; that is, we have a slack of $800
that can be used for some other purpose.
There are many types of software that can be used to solve linear programming
problems by the simplex method. Figure 3 illustrates a solution to Example 3 in
Excel, a popular spreadsheet for personal computers.
Figure 3
Matched Problem 3 Repeat Example 3 modified as follows:
Investment per Acre

Crop A Crop B Crop C Maximum Available
Seed cost $24 $40 $30 $3,600
Work days 1 2 2 160
Profit $140 $200 $160
CONCEPTUAL INSIGHT
If you are solving a system of linear equations or inequalities, then you can
multiply both sides of an equation by any nonzero number and both sides of an
inequality by any positive number without changing the solution set. This is still
the case for the simplex method, but you must be careful when you interpret the
results. For example, consider the second problem constraint in the model for
Example 3:
40x1 + 20x2 + 30x3 … 3,200
1
Multiplying both sides by 10 before introducing slack variables simplifies sub-
sequent calculations. However, performing this operation has a side effect—it
changes the units of the slack variable from dollars to tens of dollars. Compare
the following two equations:
40x1 + 20x2 + 30x3 + s2 = 3,200 s2 represents dollars
4x1 + 2x2 + 3x3 + s2= = 320 s2= represents tens of dollars
In general, if you multiply a problem constraint by a positive number, remember
to take this into account when you interpret the value of the slack variable for
that constraint.
The feasible region for the linear programming problem in Example 3 has eight
corner points, but the simplex method found the solution in only two steps. In larger
problems, the difference between the total number of corner points and the number of
steps required by the simplex method is even more dramatic. A feasible region may
have hundreds or even thousands of corner points, yet the simplex method will often
find the optimal solution in 10 or 15 steps.
To simplify this introduction to the simplex method, we have purposely avoided
certain degenerate cases that lead to difficulties. Discussion and resolution of these
problems is left to a more advanced treatment of the subject.
Exercises 5.2
A For the simplex tableaux in Problems 1–4, In Problems 9–12,
(A) Identify the basic and nonbasic variables. (A) Using slack variables, write the initial system for each linear
programming problem.
(B) Find the corresponding basic feasible solution.
(B) Write the simplex tableau, circle the first pivot, and identify
(C) Determine whether the optimal solution has been found, an
the entering and exiting variables.
additional pivot is required, or the problem has no optimal
solution. (C) Use the simplex method to solve the problem.
1. x1 x2 s1 s2 P 9. Maximize P = 15x1 + 10x2
2 1 0 3 0 12
C 3 0 1 -2 0 3 15 S
x1 + 3x2 … 10
-4 0 0 4 1 50
x1, x2 Ú 0
2. x1 x2 s1 s2 P
0 0.8 1 - 0.2 0 40
C 1 0.2 0 0.2 0 3 20 S
0 -5 0 10 1 1000 2x1 + x2 … 1000
x1, x2 Ú 0
3. x1 x2 x3 s1 s2 s3 P
-2 0 1 3 1 0 0 5 11. Repeat Problem 9 with the objective function changed to
P = 30x1 + x2.
0 1 0 -2 0 0 0 4 15
D T
-1 0 0 4 1 1 0 12 12. Repeat Problem 10 with the objective function changed to
-4 0 0 2 4 0 1 45 P = 8x1 + x2.
4. x1 x2 x3 s1 s2 s3 P B Solve the linear programming problems in Problems 13–32 using

0 2 -1 1 4 0 0 5 the simplex method.
0 1 2 0 -2 1 0 2 13. Maximize P = 30x1 + 40x2
D 4 T
1 3 0 0 5 0 0 11 subject to 2x1 + x2 … 10
0 -5 4 0 -3 0 1 27 x1 + x2 … 7
x1 + 2x2 … 12
In Problems 5–8, find the pivot element, identify the entering and
exiting variables, and perform one pivot operation. x1, x2 Ú 0
5. x1 x2 s1 s2 P 14. Maximize P = 15x1 + 20x2

1 4 1 0 0 4 subject to 2x1 + x2 … 9
C 3 5 0 1 0 3 24 S x1 + x2 … 6
-8 -5 0 0 1 0 x1 + 2x2 … 10
6. x1 x2 s1 s2 P x1, x2 Ú 0
1 1 1 0 0 300 15. Maximize P = 2x1 + 3x2
C 2 1 0 1 0 3 360 S subject to - 2x1 + x2 … 2
-5 -3 0 0 1 0 - x1 + x2 … 5
7. x1 x2 s1 s2 s3 P x2 … 6
2 1 1 0 0 0 4 x1, x2 Ú 0
3 0 1 1 0 0
D 4 8T 16. Repeat Problem 15 with P = - x1 + 3x2.
0 0 2 0 1 0 2
-4 0 -3 0 0 1 5 17. Maximize P = - x1 + 2x2
subject to - x1 + x2 … 2
8. x1 x2 s1 s2 s3 P
- x1 + 3x2 … 12
0 0 2 1 1 0 2
1 0 -4 0 1 0 4 3 x1 - 4x2 … 4
D T
0 1 5 0 2 0 11 x1, x2 Ú 0
0 0 -6 0 -5 1 18
18. Repeat Problem 17 with P = x1 + 2x2. 31. Maximize P = x1 + 2x2 + 3x3

subject to 2x1 + 2x2 + 8x3 … 600
19. Maximize P = 15x1 + 36x2
subject to x1 + 3x2 … 6 x1 + 3x2 + 2x3 … 600
x1, x2 Ú 0 3x1 + 2x2 + x3 … 400
x1, x2, x3 Ú 0
subject to 2x1 - 5x2 … 10 32. Maximize P = 10x1 + 50x2 + 10x3
x1, x2 Ú 0 subject to 3x1 + 3x2 + 3x3 … 66
6x1 - 2x2 + 4x3 … 48
21. Maximize P = 27x1 + 64x2
3x1 + 6x2 + 9x3 … 108
subject to 8x1 - 3x2 … 24
x1, x2, x3 Ú 0
x1, x2 Ú 0
22. Maximize P = 45x1 + 30x2 C In Problems 33 and 34, first solve the linear programming problem
subject to 4x1 + 3x2 … 12 by the simplex method, keeping track of the basic feasible solutions
at each step. Then graph the feasible region and illustrate the path
x1, x2 Ú 0
to the optimal solution determined by the simplex method.
23. Maximize P = 5x1 + 2x2 - x3
subject to x1 + x2 - x3 … 10
2x1 + 4x2 + 3x3 … 30
x1 + 3x2 … 48
x1, x2, x3 Ú 0
x1 + 4x2 … 60
24. Maximize P = 15x1 + 10x2 - 24x3 x2 … 14
subject to 2x1 + x2 - 2x3 … 40
x1, x2 Ú 0
2x1 + 3x2 + x3 … 80
x1, x2, x3 Ú 0
25. Maximize P = 2x1 + 3x2 + 4x3 2x1 + x2 … 28
4x1 + x2 … 42
x2 + x3 … 3
x1 … 10
x1, x2, x3 Ú 0
x1, x2 Ú 0
26. Maximize P = x1 + x2 + 2x3
subject to x1 - 2x2 + x3 … 9 Solve Problems 35 and 36 by the simplex method and also by graph-
2x1 + x2 + 2x3 … 28 ing (the geometric method). Compare and contrast the results.
x1, x2, x3 Ú 0 35. Maximize P = 2x1 + 3x2

subject to - 2x1 + x2 … 4
27. Maximize P = 3x1 + 4x2 + x3
x2 … 10
subject to x1 + 2x2 + 3x3 … 90
x1, x2 Ú 0
2x1 + x2 + x3 … 60
3x1 + x2 + 2x3 … 80 36. Maximize P = 2x1 + 3x2
subject to - x1 + x2 … 2
x1, x2, x3 Ú 0
x2 … 4
28. Maximize P = 4x1 + 2x2 + 3x3
x1, x2 Ú 0
subject to x1 + x2 + x3 … 11
2x1 + 3x2 + x3 … 20 In Problems 37–40, there is a tie for the choice of the first pivot
x1 + 3x2 + 2x3 … 20 column. Use the simplex method to solve each problem two differ-
ent ways: first by choosing column 1 as the first pivot column, and
x1, x2, x3 Ú 0
then by choosing column 2 as the first pivot column. Discuss the
29. Maximize P = 20x1 + 30x2 relationship between these two solutions.
subject to 0.6x1 + 1.2x2 … 960
37. Maximize P = x1 + x2
0.03x1 + 0.04x2 … 36
0.3x1 + 0.2x2 … 270
x1 … 6
x1, x2 Ú 0
x2 … 10
30. Repeat Problem 29 with P = 20x1 + 20x2. x1, x2 Ú 0
38. Maximize P = x1 + x2 $1,000 on daytime TV and is viewed by 14,000 potential

subject to x1 + 2x2 … 10 customers, $2,000 on prime-time TV and is viewed by
x1 … 6 24,000 potential customers, and $1,500 on late-night TV
and is viewed by 18,000 potential customers. The televi-
x2 … 4
sion station will not accept a total of more than 15 ads in all
x1, x2 Ú 0 three time periods. How many ads should be placed in each
39. Maximize P = 3x1 + 3x2 + 2x3 time period in order to maximize the number of potential
subject to x1 + x2 + 2x3 … 20 customers who will see the ads? How many potential cus-
tomers will see the ads? (Ignore repeated viewings of the ad
2x1 + x2 + 4x3 … 32
by the same potential customer.)
x1, x2, x3 Ú 0
46. Advertising. Repeat Problem 45 if the department store
40. Maximize P = 2x1 + 2x2 + x3 increases its budget to $24,000 and requires that at least half
subject to x1 + x2 + 3x3 … 10 of the ads be placed during prime-time.
2x1 + 4x2 + 5x3 … 24 47. Home construction. A contractor is planning a new housing
x1, x2, x3 Ú 0 development consisting of colonial, split-level, and ranch-
style houses. A colonial house requires 12 acre of land, $60,000
capital, and 4,000 labor-hours to construct, and returns a profit
of $20,000. A split-level house requires 12 acre of land, $60,000
Applications capital, and 3,000 labor-hours to construct, and returns a profit
of $18,000. A ranch house requires 1 acre of land, $80,000
In Problems 41–56, construct a mathematical model in the form capital, and 4,000 labor-hours to construct, and returns a profit
of a linear programming problem. (The answers in the back of of $24,000. The contractor has 30 acres of land, $3,200,000
the book for these application problems include the model.) Then capital, and 180,000 labor-hours available. How many houses
solve the problem using the simplex method. Include an interpre- of each type should be constructed to maximize the contrac-
tation of any nonzero slack variables in the optimal solution. tor’s profit? What is the maximum profit?
41. Manufacturing: resource allocation. A small company 48. Bicycle manufacturing. A company manufactures three-
manufactures three different electronic components for speed, five-speed, and ten-speed bicycles. Each bicycle
computers. Component A requires 2 hours of fabrication and passes through three departments: fabrication, painting &
1 hour of assembly; component B requires 3 hours of fabrica- plating, and final assembly. The relevant manufacturing data
tion and 1 hour of assembly; and component C requires are given in the table.
2 hours of fabrication and 2 hours of assembly. The company
has up to 1,000 labor-hours of fabrication time and 800 labor-
Labor-Hours per Bicycle Maximum
hours of assembly time available per week. The profit on
Labor-Hours
each component, A, B, and C, is $7, $8, and $10, respectively. Three- Five- Ten-
Available per
How many components of each type should the company Speed Speed Speed
Day
manufacture each week in order to maximize its profit (as-
Fabrication 3 4 5 120
suming that all components manufactured can be sold)? What
is the maximum profit? Painting & plating 5 3 5 130
Final assembly 4 3 5 120
42. Manufacturing: resource allocation. Solve Problem 41 with
Profit per bicycle ($) 80 70 100
the additional restriction that the combined total number of
components produced each week cannot exceed 420. Discuss
the effect of this restriction on the solution to Problem 41. How many bicycles of each type should the company manu-
facture per day in order to maximize its profit? What is the
43. Investment. An investor has at most $100,000 to invest in maximum profit?
government bonds, mutual funds, and money market funds.
The average yields for government bonds, mutual funds, and 49. Home building. Repeat Problem 47 if the profit on a
money market funds are 8%, 13%, and 15%, respectively. colonial house decreases from $20,000 to $17,000 and all
The investor’s policy requires that the total amount invested other data remain the same. If the slack associated with any
in mutual and money market funds not exceed the amount problem constraint is nonzero, find it.
invested in government bonds. How much should be invested 50. Bicycle manufacturing. Repeat Problem 48 if the profit
in each type of investment in order to maximize the return? on a ten-speed bicycle increases from $100 to $110 and all
What is the maximum return? other data remain the same. If the slack associated with any
44. Investment. Repeat Problem 43 under the additional as- problem constraint is nonzero, find it.
sumption that no more than $30,000 can be invested in 51. Home building. Repeat Problem 47 if the profit on a
money market funds. colonial house increases from $20,000 to $25,000 and all
45. Advertising. A department store has up to $20,000 to other data remain the same. If the slack associated with any
spend on television advertising for a sale. All ads will be problem constraint is nonzero, find it.
placed with one television station. A 30-second ad costs
52. Bicycle manufacturing. Repeat Problem 48 if the profit on 55. Opinion survey. A political scientist received a grant to
a five-speed bicycle increases from $70 to $110 and all other fund a research project on voting trends. The budget includes
data remain the same. If the slack associated with any prob- $3,200 for conducting door-to-door interviews on the day be-
lem constraint is nonzero, find it. fore an election. Undergraduate students, graduate students,
and faculty members will be hired to conduct the interviews.
53. Animal nutrition. The natural diet of a certain animal
Each undergraduate student will conduct 18 interviews for
consists of three foods: A, B, and C. The number of units of
$100. Each graduate student will conduct 25 interviews
calcium, iron, and protein in 1 gram of each food and the
for $150. Each faculty member will conduct 30 interviews
average daily intake are given in the table. A scientist wants
for $200. Due to limited transportation facilities, no more
to investigate the effect of increasing the protein in the ani-
than 20 interviewers can be hired. How many undergraduate
mal’s diet while not allowing the units of calcium and iron to
students, graduate students, and faculty members should be
exceed its average daily intakes. How many grams of each
hired in order to maximize the number of interviews? What is
food should be used to maximize the amount of protein in the
the maximum number of interviews?
diet? What is the maximum amount of protein?
56. Opinion survey. Repeat Problem 55 if one of the require-
Units per Gram ments of the grant is that at least 50% of the interviewers be
Average Daily undergraduate students.
Food A Food B Food C
Intake (units)
Calcium 1 3 2 30 Answers to Matched Problems
Iron 2 1 2 24
1. Max P = 6 when x1 = 1 and x2 = 4
Protein 3 4 5 60
2. No optimal solution
54. Animal nutrition. Repeat Problem 53 if the scientist wants 3. 40 acres of crop A, 60 acres of crop B, no crop C; max
to maximize the daily calcium intake while not allowing the P = $17,600 (since s2 = 240, $240 out of the $3,600 will
intake of iron or protein to exceed the average daily intake. not be spent).
5.3 The Dual Problem:

Minimization with Problem Constraints of the Form Ú
■■ Formation of the Dual Problem In the preceding section, we restricted attention to standard maximization problems
■■ Solution of Minimization Problems
(problem constraints of the form … , with nonnegative constants on the right and any
real numbers as objective function coefficients). Now we will consider minimization
■■ Application: Transportation Problem problems with Ú problem constraints. These two types of problems turn out to be
■■ Summary of Problem Types very closely related.
and Solution Methods
Formation of the Dual Problem
Associated with each minimization problem with Ú constraints is a maximization
problem called the dual problem. To illustrate the procedure for forming the dual
problem, consider the following minimization problem:
Minimize C = 16x1 + 45x2
subject to 2x1 + 5x2 Ú 50
(1)
x1 + 3x2 Ú 27
x1, x2 Ú 0
The first step in forming the dual problem is to construct a matrix using the prob-
lem constraints and the objective function written in the following form:
2x1 + 5x2 Ú 50 2 5 50
x1 + 3x2 Ú 27 A = C 1 3 3 27 S
16x1 + 45x2 = C 16 45 1
SECTION 5.3 The Dual Problem: Minimization with Problem Constraints of the Form Ú 295
Do not confuse matrix A with the simplex tableau. We use a solid

! CAUTION horizontal line in matrix A to help distinguish the dual matrix
from the simplex tableau. No slack variables are involved in
matrix A, and the coefficient of C is in the same column as the constants from the
problem constraints.
Now we will form a second matrix called the transpose of A. In general, the
transpose of a given matrix A is the matrix AT formed by interchanging the rows and
corresponding columns of A (first row with first column, second row with second
column, and so on).
2 5 50 R1 in A = C1 in AT
A = £ 1 3 3 27 § R2 in A = C2 in AT
16 45 1 R3 in A = C3 in AT n n n
2 1 16
AT = £ 5 3 3 45 § AT is the transpose of A.
50 27 1
We can use the rows of AT to define a new linear programming problem. This
new problem will always be a maximization problem with … problem constraints. To
avoid confusion, we will use different variables in this new problem:
y1 y2
2y1 + y2 … 16 2 1 16
5y1 + 3y2 … 45 T
A = C 5 3 3 45 S
50y1 + 27y2 = P 50 27 1
The dual of the minimization problem (1) is the following maximization problem:
Maximize P = 50y1 + 27y2
subject to 2y1 + y2 … 16
(2)
5y1 + 3y2 … 45
y1, y2 Ú 0
Excluding the nonnegative constraints, the components of a linear programming prob-

lem can be divided into three categories: the coefficients of the objective function, the
coefficients of the problem constraints, and the constants on the right side of the prob-
lem constraints. Write a verbal description of the relationship between the components
of the original minimization problem (1) and the dual maximization problem (2).
The procedure for forming the dual problem is summarized in the following box:
PROCEDURE Formation of the Dual Problem

Given a minimization problem with Ú problem constraints,
Step 1 Use the coefficients and constants in the problem constraints and the objec-
tive function to form a matrix A with the coefficients of the objective function
in the last row.
Step 2 Interchange the rows and columns of matrix A to form the matrix AT, the
transpose of A.
Step 3 Use the rows of AT to form a maximization problem with … problem
constraints.
EXAMPLE 1 Forming the Dual Problem Form the dual problem:
Minimize C = 40x1 + 12x2 + 40x3

subject to 2x1 + x2 + 5x3 Ú 20
4x1 + x2 + x3 Ú 30
x1 , x2, x3 Ú 0
SOLUTION
Step 1 Form the matrix A:
2 1 5 20
A = C 4 1 1 3 30 S
40 12 40 1
Step 2 Form the matrix AT, the transpose of A:
2 4 40
1 1 4 12
AT = D T
5 1 40
20 30 1
Step 3 State the dual problem:

subject to 2y1 + 4y2 … 40
y1 + y2 … 12
5y1 + y2 … 40
y1, y2 Ú 0
Matched Problem 1 Form the dual problem:

Minimize C = 16x1 + 9x2 + 21x3
subject to x1 + x2 + 3x3 Ú 12
2x1 + x2 + x3 Ú 16
x1, x2, x3 Ú 0
Solution of Minimization Problems

The following theorem establishes the relationship between the solution of a minimi-
zation problem and the solution of its dual problem:
THEOREM 1 Fundamental Principle of Duality

A minimization problem has a solution if and only if its dual problem has a solution.
If a solution exists, then the optimal value of the minimization problem is the same
as the optimal value of the dual problem.
The proof of Theorem 1 is beyond the scope of this text. However, we can
illustrate Theorem 1 by solving minimization problem (1) and its dual maximization
problem (2) geometrically.
Original Problem (1) Dual Problem (2)

Minimize C = 16x1 + 45x2 Maximize P = 50y1 + 27y2
subject to 2x1 + 5x2 Ú 50 subject to 2y1 + y2 … 16
x1 + 3x2 Ú 27 5y1 + 3y2 … 45
x1, x2 Ú 0 y1, y2 Ú 0
x2 y2
30 15 (0, 15)
20 10 (3, 10)
(0, 10)
10 5
(15, 4)
(27, 0) (8, 0)
x1 y1
10 20 30 (0, 0) 5 10 15
Corner Point Corner Point

1x1, x2 2 C = 16x1 + 45x2 1y1, y2 2 P = 50y1 + 27y2
10, 102 450 10, 02 0
115, 42 420 10, 152 405
127, 02 432 13, 102 420
Min C = 420 at 115, 42 18, 02 400
Max P = 420 at 13, 102
Note that the minimum value of C in problem (1) is the same as the maximum
value of P in problem (2). The optimal solutions producing this optimal value are
different: 115, 42 is the optimal solution for problem (1), and 13, 102 is the optimal
solution for problem (2). Theorem 1 only guarantees that the optimal values of a mini-
mization problem and its dual are equal, not that the optimal solutions are the same. In
general, it is not possible to determine an optimal solution for a minimization problem
by examining the feasible set for the dual problem. However, it is possible to apply
the simplex method to the dual problem and find both the optimal value and an opti-
mal solution to the original minimization problem. To see how this is done, we will
solve problem (2) using the simplex method.
For reasons that will become clear later, we will use the variables x1 and x2 from
the original problem as the slack variables in the dual problem:
2y1 + y2 + x1 = 16
5y1 + 3y2 + x2 = 45 Initial system for the dual problem
- 50y1 - 27y2 + P = 0
y1 y2 x1 x2 P
x1 2 1 1 0 0 16 0.5R1 S R1
x2 £ 5 3 0 1 0 3 45 §
P - 50 - 27 0 0 1 0
1 0.5 0.5 0 0 8
∼£ 5 3 0 1 0 3 45 § 1 - 52R1 + R2 S R2
- 50 - 27 0 0 1 0 50R1 + R3 S R3
y1 1 0.5 0.5 0 0 8
∼ x2 £ 0 0.5 - 2.5 1 0 3 5§ 2R2 S R2
P 0 -2 25 0 1 400
1 0.5 0.5 0 0 8 1 - 0.52R2 + R1 S R1

∼£ 0 1 -5 2 0 3 10 §
0 -2 25 0 1 400 2R2 + R3 S R3
y1 1 0 3 -1 0 3
∼ y2 £ 0 1 -5 2 0 3 10 §
P 0 0 15 4 1 420
Since all indicators in the bottom row are nonnegative, the solution to the dual problem is
y1 = 3, y2 = 10, x1 = 0, x2 = 0, P = 420
which agrees with our earlier geometric solution. Furthermore, examining the bottom
row of the final simplex tableau, we see the same optimal solution to the minimiza-
tion problem that we obtained directly by the geometric method:
Min C = 420 at x1 = 15, x2 = 4
This is no accident.
An optimal solution to a minimization problem always can be obtained
from the bottom row of the final simplex tableau for the dual problem.
We can see that using x1 and x2 as slack variables in the dual problem makes it
easy to identify the solution of the original problem.
The simplex method can be used to solve any standard maximization problem.
Which of the following minimization problems have dual problems that are standard
maximization problems? (Do not solve the problems.)
(A) Minimize C = 2x1 + 3x2
subject to 2x1 - 5x2 Ú 4
x1 - 3x2 Ú -6
x1, x2 Ú 0
(B) Minimize C = 2x1 - 3x2
subject to - 2x1 + 5x2 Ú 4
- x1 + 3x2 Ú 6
x1, x2 Ú 0
What conditions must a minimization problem satisfy so that its dual problem is a
standard maximization problem?
The procedure for solving a minimization problem by applying the simplex

method to its dual problem is summarized in the following box:
PROCEDURE Solution of a Minimization Problem

Given a minimization problem with nonnegative coefficients in the objective function,
Step 1 Write all problem constraints as Ú inequalities. (This may introduce negative
numbers on the right side of some problem constraints.)
Step 2 Form the dual problem.
Step 3 Write the initial system of the dual problem, using the variables from the
minimization problem as slack variables.
Step 4 Use the simplex method to solve the dual problem.

Step 5 Read the solution of the minimization problem from the bottom row of the
final simplex tableau in step 4.
Note: If the dual problem has no optimal solution, the minimization problem has
no optimal solution.
EXAMPLE 2 Solving a Minimization Problem Solve the following minimization problem by

maximizing the dual problem:
Minimize C = 40x1 + 12x2 + 40x3
4x1 + x2 + x3 Ú 30
x1, x2, x3 Ú 0
SOLUTION From Example 1, the dual problem is

y1 + y2 … 12
5y1 + y2 … 40
y1, y2 Ú 0
Using x1, x2, and x3 for slack variables, we obtain the initial system for the dual
problem:
2y1 + 4y2 + x1 = 40
y1 + y2 + x2 = 12
5y1 + y2 + x3 = 40
- 20y1 - 30y2 + P = 0
Now we form the simplex tableau and solve the dual problem:
y1 y2 x1 x2 x3 P
1
x1 2 4 1 0 0 0 40 4 R1 S R1
x2 1 1 0 1 0 0 12
≥ 4 ¥
x3 5 1 0 0 1 0 40
P - 20 - 30 0 0 0 1 0
1 1
2 1 4 0 0 0 10
1 1 0 1 0 0 12 1 - 12R1 + R2 S R2
∼≥ 4 ¥
5 1 0 0 1 0 40 1 - 12R1 + R3 S R3
- 20 - 30 0 0 0 1 0 30R1 + R4 S R4
1 1
y2 2 1 4 0 0 0 10
1
x2 2 0 - 14 1 0 0 2 2R2 S R2
∼ E 5 U
9
x3 2 0 - 14 0 1 0 30
15
P -5 0 2 0 0 1 300
1 1
2 1 4 0 0 0 10 1 - 12 2 R2 + R1 S R1
1 0 - 12 2 0 0 4
∼E 5 U
9
2 0 - 14 0 1 0 30 1 - 92 2 R2 + R3 S R3
15
-5 0 2 0 0 1 300 5R2 + R4 S R4
1
y2 0 1 2 -1 0 0 8
1
y1 1 0 - 2 2 0 0 4 4
∼ ≥ ¥
x3 0 0 2 -9 1 0 12
P 0 0 5 10 0 1 320
From the bottom row of this tableau, we see that
Min C = 320 at x1 = 5, x2 = 10, x3 = 0
Matched Problem 2 Solve the following minimization problem by maximiz-

ing the dual problem (see Matched Problem 1):
Minimize C = 16x1 + 9x2 + 21x3
subject to x1 + x2 + 3x3 Ú 12
2x1 + x2 + x3 Ú 16
x1, x2, x3 Ú 0
CONCEPTUAL INSIGHT
In Section 5.2, we noted that multiplying a problem constraint by a number
changes the units of the slack variable. This requires special interpretation of the
value of the slack variable in the optimal solution, but causes no serious prob-
lems. However, when using the dual method, multiplying a problem constraint in
the dual problem by a number can have serious consequences—the bottom row
of the final simplex tableau may no longer give the correct solution to the mini-
mization problem. To see this, refer to the first problem constraint of the dual
problem in Example 2:
2y1 + 4y2 … 40
If we multiply this constraint by 12 and then solve, the final tableau is:
y1 y2 x1 x2 x3 P
0 1 1 -1 0 0 8
1 0 -1 2 0 0
D 4 4T
0 0 4 -9 1 0 12
0 0 10 10 0 1 320
The bottom row of this tableau indicates that the optimal solution to the minimiza-
tion problem is C = 320 at x1 = 10 and x2 = 10. This is not the correct answer
(x1 = 5 is the correct answer). Thus, you should never multiply a problem
constraint in a maximization problem by a number if that maximization
problem is being used to solve a minimization problem. You may still simplify
problem constraints in a minimization problem before forming the dual problem.
EXAMPLE 3 Solving a Minimization Problem Solve the following minimization problem by

maximizing the dual problem:
subject to x1 - x2 Ú 1
- x1 + x2 Ú 2
x1, x2 Ú 0
1 -1 1 1 -1 5
SOLUTION A = C -1 1 3 2S T
A = C -1 1 3 10 S
5 10 1 1 2 1
The dual problem is
Maximize P = y1 + 2y2
subject to y1 - y2 … 5
- y1 + y2 … 10
y1, y2 Ú 0
Introduce slack variables x1 and x2, and form the initial system for the dual problem:
y1 - y2 + x1 = 5
- y1 + y2 + x2 = 10
- y1 - 2y2 + P = 0
Form the simplex tableau and solve:
x2 y1 y2 x1 x2 P
5 x1 1 -1 1 0 0 5 R2 + R1 S R1
A x2 C - 1 1 0 1 0 3 10 S
2x1 1 x2 $ 2 P -1 -2 0 0 1 0 2R2 + R3 S R3
B
0 0 1 1 0 15 No positive elements
x1
25 5 ∼ C - 1 1 0 1 0 3 10 S above dashed line in
-3 0 0 2 1 20 pivot column
x1 2 x2 $ 1 c
2x1 1 x2 5 2 Pivot column
25 The - 3 in the bottom row indicates that column 1 is the pivot column. Since no
x1 2 x2 5 1 A>B5[ positive elements appear in the pivot column above the dashed line, we are unable to
Figure 1 select a pivot row. We stop the pivot operation and conclude that this maximization
problem has no optimal solution (see Fig. 2, Section 5.2). Theorem 1 now implies
that the original minimization problem has no solution. The graph of the inequalities
in the minimization problem (Fig. 1) shows that the feasible region is empty; so it is
not surprising that an optimal solution does not exist.
Matched Problem 3 Solve the following minimization problem by maximiz-

ing the dual problem:
subject to x1 - 2x2 Ú 2
- x1 + x2 Ú 1
x1, x2 Ú 0
Application: Transportation Problem
One of the first applications of linear programming was to minimize the cost of trans-
porting materials. Problems of this type are referred to as transportation problems.
EXAMPLE 4 Transportation Problem A computer manufacturing company has two assembly

plants, plant A and plant B, and two distribution outlets, outlet I and outlet II. Plant A
can assemble at most 700 computers a month, and plant B can assemble at most 900
computers a month. Outlet I must have at least 500 computers a month, and outlet II
must have at least 1,000 computers a month. Transportation costs for shipping one
computer from each plant to each outlet are as follows: $6 from plant A to outlet I,
$5 from plant A to outlet II, $4 from plant B to outlet I, $8 from plant B to outlet II.
Find a shipping schedule that minimizes the total cost of shipping the computers
from the assembly plants to the distribution outlets. What is this minimum cost?
Outlet I
SOLUTION To form a shipping schedule, we must decide how many computers to
Plant A ship from either plant to either outlet (Fig. 2). This will involve four decision variables:
From A to I
Fr
x1 = number of computers shipped from plant A to outlet I
om I
to
B x2 = number of computers shipped from plant A to outlet II
A
om to
Fr II x3 = number of computers shipped from plant B to outlet I
From B to II
x4 = number of computers shipped from plant B to outlet II
Plant B Outlet II Next, we summarize the relevant data in a table. Note that we do not follow the usu-
Figure 2 al technique of associating each variable with a column of the table. Instead, sources
are associated with the rows, and destinations are associated with the columns.
Distribution Outlet Assembly

I II Capacity
Plant A $6 $5 700
Plant B $4 $8 900
Minimum required 500 1,000
The total number of computers shipped from plant A is x1 + x2. Since this cannot
exceed the assembly capacity at A, we have
x1 + x2 … 700 Number shipped from plant A
Similarly, the total number shipped from plant B must satisfy
x3 + x4 … 900 Number shipped from plant B
The total number shipped to each outlet must satisfy
x1 + x3 Ú 500 Number shipped to outlet I
and
x2 + x4 Ú 1,000 Number shipped to outlet II
Using the shipping charges in the table, the total shipping charges are
C = 6x1 + 5x2 + 4x3 + 8x4
We must solve the following linear programming problem:
Minimize C = 6x1 + 5x2 + 4x3 + 8x4
subject to x1 + x2 … 700 Available from A
x3 + x4 … 900 Available from B
x1 + x3 Ú 500 Required at I
x2 + x4 Ú 1,000 Required at II
x1, x2, x3, x4 Ú 0
Before we can solve this problem, we must multiply the first two constraints
by - 1 so that all the problem constraints are of the Ú type. This will introduce
negative constants into the minimization problem but not into the dual problem.
Since the coefficients of C are nonnegative, the constants in the dual problem will
be nonnegative and the dual will be a standard maximization problem. The problem
can now be stated as
Minimize C = 6x1 + 5x2 + 4x3 + 8x4
subject to - x1 - x2 Ú - 700
- x3 - x4 Ú - 900
x1 + x3 Ú 500
x2 + x4 Ú 1,000
x1, x2, x3, x4 Ú 0
-1 -1 0 0 - 700
0 0 -1 -1 - 900
A = E 1 0 1 0 5 500 U
0 1 0 1 1,000
6 5 4 8 1
-1 0 1 0 6
-1 0 0 1 5
T
A = E 0 -1 1 0 5 4U
0 -1 0 1 8
- 700 - 900 500 1,000 1
The dual problem is
Maximize P = - 700y1 - 900y2 + 500y3 + 1,000y4
subject to - y1 + y3 … 6
- y1 + y4 … 5
- y2 + y3 … 4
- y2 + y4 … 8
y1, y2, y3, y4 Ú 0
Introduce slack variables x1, x2, x3, and x4, and form the initial system for the dual
problem:
- y1 + y3 + x1 = 6
- y1 + y4 + x2 = 5
- y2 + y3 + x3 = 4
- y2 + y4 + x4 = 8
700y1 + 900y2 - 500y3 - 1,000y4 + P = 0
Form the simplex tableau and solve:
y1 y2 y3 y4 x1 x2 x3 x4 P
x1 -1 0 1 0 1 0 0 0 0 6
x2 -1 0 0 1 0 1 0 0 0 5
x3 E 0 -1 1 0 0 0 1 0 0 5 4U
x4 0 -1 0 1 0 0 0 1 0 8 1 - 12R2 + R4 S R4
P 700 900 - 500 - 1,000 0 0 0 0 1 0 1,000R2 + R5 S R5
y1 y2 y3 y4 x1 x2 x3 x4 P
x1 -1 0 1 0 1 0 0 0 0 6 1 - 12R3 + R1 S R1
y4 -1 0 0 1 0 1 0 0 0 5
∼ x3 E 0 -1 1 0 0 0 1 0 0 5 4U
x4 1 -1 0 0 0 -1 0 1 0 3
P - 300 900 - 500 0 0 1,000 0 0 1 5,000 500R3 + R5 S R5
x1 -1 1 0 0 1 0 -1 0 0 2 R4 + R1 S R1
y4 -1 0 0 1 0 1 0 0 0 5 R4 + R2 S R2
∼ y3 E 0 -1 1 0 0 0 1 0 0 5 4U
x4 1 -1 0 0 0 -1 0 1 0 3
P - 300 400 0 0 0 1,000 500 0 1 7,000 300R4 + R5 S R5
x1 0 0 0 0 1 -1 -1 1 0 5
y4 0 -1 0 1 0 0 0 1 0 8
∼ y3 E 0 -1 1 0 0 0 1 0 0 5 4U
y1 1 -1 0 0 0 -1 0 1 0 3
P 0 100 0 0 0 700 500 300 1 7,900
From the bottom row of this tableau, we have

Min C = 7,900 at x1 = 0, x2 = 700, x3 = 500, x4 = 300
The shipping schedule that minimizes the shipping charges is 700 from plant A to
outlet II, 500 from plant B to outlet I, and 300 from plant B to outlet II. The total
shipping cost is $7,900.
Figure 3 shows a solution to Example 4 in Excel. Notice that Excel permits the
user to organize the original data and the solution in a format that is clear and easy
to read. This is one of the main advantages of using spreadsheets to solve linear
programming problems.
Figure 3
Matched Problem 4 Repeat Example 4 if the shipping charge from plant A

to outlet I is increased to $7 and the shipping charge from plant B to outlet II is
decreased to $3.
Summary of Problem Types and Solution Methods

In this and the preceding sections, we have solved both maximization and minimi-
zation problems, but with certain restrictions on the problem constraints, constants
on the right, and/or objective function coefficients. Table 1 summarizes the types of
problems and methods of solution we have considered so far.
Table 1 Summary of Problem Types and Simplex Solution Methods

Problem Right-Side Coefficients of
Problem Type Constraints Constants Objective Function Method of Solution
1. Maximization … Nonnegative Any real numbers Simplex method with
slack variables
2. Minimization Ú Any real numbers Nonnegative Form dual problem and
solve by simplex method
with slack variables
The next section develops a generalized version of the simplex method that can
handle both maximization and minimization problems with any combination of … , Ú ,
and = problem constraints.
Exercises 5.3
In Problems 1–8, find the transpose of each matrix. (B) Find the optimal solution of the minimization problem.
A 1. -5 3 2. 4 -1 11. Minimize C = 21x1 + 50x2
c d c d
2 7 1 6 subject to 2x1 + 5x2 Ú 12
3. 1 7 8 4. 9 1 3 3x1 + 7x2 Ú 17
C6 3 5S C1 8 2S x1, x2 Ú 0
9 4 2 3 2 7 Maximize P = 12y1 + 17y2
5. 2 1 -6 0 -1 6. 7 3 -1 3 5y1 + 7y2 … 50
c d c d
5 2 0 1 3 -6 1 0 -9 y1 , y2 Ú 0
7. 2 3 0 -1 8. 1 -1 3 2 y1 y2 x1 x2 P
£ -5 1 8 4§ 1 -4 0 2 0 1 5 -2 0 5
E 4 -5 6 1U C 1 0 -7 3 0 3 3S
6 9 0 -7
-3 8 0 -1 0 0 1 2 1 121
2 7 -3 1
12. Minimize C = 50x1 + 70x2
In Problems 9 and 10, subject to 2x1 + x2 Ú 4
(A) Form the dual problem. x1 + 2x2 Ú 5
x1, x2 Ú 0
(B) Write the initial system for the dual problem.
(C) Write the initial simplex tableau for the dual problem and Maximize P = 4y1 + 5y2
label the columns of the tableau. subject to 2y1 + y2 … 50
y1 + 2y2 … 70
9. Minimize C = 8x1 + 9x2 y1, y2 Ú 0
subject to x1 + 3x2 Ú 4
y1 y2 x1 x2 P
2x1 + x2 Ú 5 2
1 0 3 - 13 0 10
x1, x2 Ú 0 1 2 3
C0 1 - 3 3 0 30 S
10. Minimize C = 4x1 + 3x2 0 0 1 2 1 190
subject to - 3x1 + 2x2 Ú 10
In Problems 13–20,
x1 + x2 Ú 6
x1, x2 Ú 0 (A) Form the dual problem.
In Problems 11 and 12, a minimization problem, the correspond- (B) Find the solution to the original problem by applying the sim-
ing dual problem, and the final simplex tableau in the solution of plex method to the dual problem.
the dual problem are given.
13. Minimize C = 9x1 + 2x2
(A) Find the optimal solution of the dual problem. subject to 4x1 + x2 Ú 13
3x1 + x2 Ú 12
x1, x2 Ú 0
14. Minimize C = x1 + 4x2 25. Minimize C = 10x1 + 30x2

subject to x1 + 2x2 Ú 5 subject to 2x1 + x2 Ú 16
x1 + 3x2 Ú 6 x1 + x2 Ú 12
x1, x2 Ú 0 x1 + 2x2 Ú 14
x1, x2 Ú 0
subject to 2x1 + 3x2 Ú 15 26. Minimize C = 6x1 + 10x2
x1 + 2x2 Ú 8 x1 + 2x2 Ú 10
x1, x2 Ú 0 x1 + x2 Ú 6
3x1 + x2 Ú 8
16. Minimize C = 3x1 + 5x2 x1, x2 Ú 0
subject to 2x1 + 3x2 Ú 7 27. Minimize C = 5x1 + 7x2
x1 + 2x2 Ú 4 subject to x1 Ú 4
x1, x2 Ú 0 x1 + x2 Ú 8
17. Minimize C = 11x1 + 4x2 x1 + 2x2 Ú 10
subject to 2x1 + x2 Ú 8 x1, x2 Ú 0
- 2x1 + 3x2 Ú 4 28. Minimize C = 4x1 + 5x2
x1, x2 Ú 0 subject to 2x1 + x2 Ú 12
x1 + x2 Ú 9
x2 Ú 4
2x1 + x2 Ú 6
x1, x2 Ú 0
x1 + x2 Ú 4
x1, x2 Ú 0 29. Minimize C = 60x1 + 25x2
subject to 2x1 + x2 Ú 4
19. Minimize C = 7x1 + 9x2 x1, x2 Ú 0
subject to - 3x1 + x2 Ú 6
30. Minimize C = 35x1 + 90x2
x1 - 2x2 Ú 4
x1, x2 Ú 0
x1, x2 Ú 0
20. Minimize C = 10x1 + 15x2 31. Minimize C = 10x1 + 25x2 + 12x3
subject to - 4x1 + x2 Ú 12 subject to 2x1 + 6x2 + 3x3 Ú 6
12x1 - 3x2 Ú 10 x1, x2, x3 Ú 0
x1, x2 Ú 0 32. Minimize C = 10x1 + 5x2 + 15x3
subject to 7x1 + 3x2 + 6x3 Ú 42
B Solve the linear programming problems in Problems 21–36
by applying the simplex method to the dual problem. x1, x2, x3 Ú 0
21. Minimize C = 3x1 + 9x2 33. Minimize C = 2x1 + 9x2 + x3

subject to 2x1 + x2 Ú 8 subject to x1 + 4x2 + 2x3 Ú 5
x1 + 2x2 Ú 8 3x1 + x2 + 2x3 Ú 4
x1, x2 Ú 0 x1, x2, x3 Ú 0
22. Minimize C = 2x1 + x2 34. Minimize C = 14x1 + 8x2 + 20x3
subject to x1 + x2 Ú 8 subject to x1 + x2 + 3x3 Ú 6
x1 + 2x2 Ú 4 2x1 + x2 + x3 Ú 9
x1, x2 Ú 0 x1, x2, x3 Ú 0
23. Minimize C = 7x1 + 5x2 35. Minimize C = 5x1 + 2x2 + 2x3
subject to x1 + x2 Ú 4 subject to x1 - 4x2 + x3 Ú 6
x1 - 2x2 Ú - 8 - x1 + x2 - 2x3 Ú 4
- 2x1 + x2 Ú - 8 x1, x2, x3 Ú 0
x1, x2 Ú 0 36. Minimize C = 6x1 + 8x2 + 3x3
24. Minimize C = 10x1 + 4x2 subject to - 3x1 - 2x2 + x3 Ú 4
subject to 2x1 + x2 Ú 6 x1 + x2 - x3 Ú 2
x1 - 4x2 Ú - 24 x1, x2, x3 Ú 0
- 8x1 + 5x2 Ú - 24
x1, x2 Ú 0
37. A minimization problem has 4 variables and 2 problem con- 46. Minimize C = 6x1 + 8x2 + 12x3
straints. How many variables and problem constraints are in subject to x1 + 3x2 + 3x3 Ú 6
the dual problem? x1 + 5x2 + 5x3 Ú 4
38. A minimization problem has 3 variables and 5 problem con- 2x1 + 2x2 + 3x3 Ú 8
straints. How many variables and problem constraints are in x1, x2, x3 Ú 0
the dual problem?
47. Minimize C = 5x1 + 4x2 + 5x3 + 6x4
39. If you want to solve a minimization problem by applying the
geometric method to the dual problem, how many variables
x3 + x4 … 25
and problem constraints must be in the original problem?
x1 + x3 Ú 20
40. If you want to solve a minimization problem by applying x2 + x4 Ú 15
the geometric method to the original problem, how many x1, x2, x3, x4 Ú 0
variables and problem constraints must be in the original
48. Repeat Problem 47 with C = 4x1 + 7x2 + 5x3 + 6x4.
problem?
In Problems 41 and 42,

(A) Form the dual problem.
Applications
(B) Is the dual problem a standard maximization problem in stan-
dard form? Explain. In Problems 49–58, construct a mathematical model in the form
41. Minimize C = 4x1 - x2 of a linear programming problem. (The answers in the back of
subject to 5x1 + 2x2 Ú 7 the book for these application problems include the model.) Then
4x1 + 6x2 Ú 10 solve the problem by applying the simplex method to the dual
problem.
x1, x2 Ú 0
42. Minimize C = - 2x1 + 9x2 49. Breads. A bakery bakes brown and multigrain breads in
three ovens. Per hour of operation, Oven A bakes 15 loaves
subject to 3x1 - x2 Ú - 8
of brown bread and 30 loaves of multigrain bread. Oven B
x1 + 4x2 Ú 4 bakes 15 loaves of brown and 15 loaves of multigrain breads,
x1, x2 Ú 0 and Oven C bakes 30 loaves of brown and 15 loaves of mul-
In Problems 43 and 44, tigrain breads. It costs $8 per hour to operate Oven A, $6 per
hour to operate Oven B, and $20 per hour to operate Oven C.
(A) Form an equivalent minimization problem with Ú problem The bakery needs to bake at least 105 loaves of brown bread
constraints (multiply inequalities by - 1 if necessary), and at least 60 loaves of multigrain bread per day. How many
(B) Form the dual of the equivalent problem. hours per day should each oven operate in order to bake the
required amount of breads and minimize the cost of baking?
(C) Is the dual problem a standard maximization problem in stan- What is the minimum baking cost?
dard form? Explain.
50. Refining. An oil refining company operates two refineries,
43. Minimize C = 3x1 + x2 + 5x3 each refining three types of oil. The sea-shore refinery can
subject to 2x1 - 6x2 - x3 … 10 refine 2 tons of very light oil, 1 ton of light oil, and 1 ton of
- 5x1 + x2 + 4x3 Ú 15 medium oil in one hour of operation. The in-sea refinery can
x1, x2, x3 Ú 0 refine 1 ton of very light oil, 1 ton of light oil, and 3 tons of
medium oil. To satisfy existing orders, the company needs to
44. Minimize C = 25x1 + 30x2 + 50x3 refine 120 tons of very light oil, 66 tons of light oil, and 180
subject to - x1 + 3x2 - 8x3 … 12 tons of medium oil. The cost of operating each refinery varies,
6x1 - 4x2 + 5x3 … 20 depending on the conditions while refining the oil. If it costs
x1, x2, x3 Ú 0 $9 per hour to operate the sea-shore refinery and $6 per hour to
operate the in-sea refinery, how many hours should each refin-
C Solve the linear programming problems in Problems 45–48 by ap- ery operate to refine the required quantities of oil and minimize
plying the simplex method to the dual problem. the cost of refining? What is the minimum refining cost?
45. Minimize C = 16x1 + 8x2 + 4x3 51. Breads. Repeat Problem 49 if the demand for multigrain
subject to 3x1 + 2x2 + 2x3 Ú 16 bread increases from 60 loaves to 150 loaves per day and all
4x1 + 3x2 + x3 Ú 14 other data remain the same.
5x1 + 3x2 + x3 Ú 12
x1, x2, x3 Ú 0
52. Refining. Repeat Problem 50 if it costs $3 per hour to oper- To balance the human resources, the management de-
ate the sea-shore refinery and $12 per hour to operate the cides to provide cab service to a few employees from the
in-sea refinery and all other data remain the same. overstaffed branches so that they can work in the under-
staffed branches. East branch has 50 more employees than
53. Breads. Repeat Problem 49 if the demand for multigrain normal, and West branch has 30 more employees than
bread increases from 60 loaves to 180 loaves per day and all normal. South branch can engage 70 more employees, and
other data remain the same. North branch can engage 20 more employees. The weekly
54. Refining Repeat Problem 50 if it costs $15 per hour to oper- cost of providing cab to an employee from East branch to
ate the sea-shore refinery and $15 per hour to operate the South branch is $10, from East branch to North branch
in-sea refinery and all other data remain the same. is $8, from West branch to South branch is $6, and from
West branch to North branch is $2. Determine the number
55. Human nutrition. A dietitian arranges a special diet using three
of employees from each overstaffed branch who should be
foods: L, M, and N. Each ounce of food L contains 20 units of
provided the cab service so that they can work in each un-
calcium, 10 units of iron, 10 units of vitamin A, and 20 units of
derstaffed branch in order to balance the human resources
cholesterol. Each ounce of food M contains 10 units of calcium,
and minimize the cost of providing cab to employees.
10 units of iron, 15 units of vitamin A, and 24 units of choles-
What is the minimum cost?
terol. Each ounce of food N contains 10 units of calcium, 10
units of iron, 10 units of vitamin A, and 18 units of cholesterol. 58. Employment: resources allocation. Repeat Problem 57 if the
If the minimum daily requirements are 300 units of calcium, weekly cost of providing cab service to an employee from West
200 units of iron, and 240 units of vitamin A, how many ounces branch to North is $12 and all other data remain the same.
of each food should be used to meet the minimum requirements
and simultaneously minimize the cholesterol intake? What is the
minimum cholesterol intake? Answers to Matched Problems
56. Plant food. A farmer can buy three types of plant food: mix A, 1. Maximize P = 12y1 + 16y2
mix B, and mix C. Each cubic yard of mix A contains 20 pounds subject to y1 + 2y2 … 16
of phosphoric acid, 10 pounds of nitrogen, and 10 pounds of y1 + y2 … 9
potash. Each cubic yard of mix B contains 10 pounds of phos-
3y1 + y2 … 21
phoric acid, 10 pounds of nitrogen, and 15 pounds of potash.
Each cubic yard of mix C contains 20 pounds of phosphoric y1, y2 Ú 0
acid, 20 pounds of nitrogen, and 5 pounds of potash. The 2. Min C = 136 at x1 = 4, x2 = 8, x3 = 0
minimum monthly requirements are 480 pounds of phosphoric 3. Dual problem:
acid, 320 pounds of nitrogen, and 225 pounds of potash. If mix Maximize P = 2y1 + y2
A costs $30 per cubic yard, mix B costs $36 per cubic yard, and subject to y1 - y2 … 2
mix C costs $39 per cubic yard, how many cubic yards of each - 2y1 + y2 … 3
mix should the farmer blend to meet the minimum monthly y1, y2 Ú 0
requirements at a minimal cost? What is the minimum cost? No optimal solution
57. Employment: resource allocation. A corporate house has 4. 600 from plant A to outlet II, 500 from plant B to outlet I, 400
two overstaffed branches and two understaffed branches. from plant B to outlet II; total shipping cost is $6,200.
5.4 Maximization and Minimization with

Mixed Problem Constraints
■■ Introduction to the Big M Method In the preceding two sections, we have seen how to solve both maximization and
■■ Big M Method
minimization problems, but with rather severe restrictions on problem constraints,
right-side constants, and/or objective function coefficients (see the summary in Table 1 of
■■ Minimization by the Big M Method Section 5.3). In this section we present a generalized version of the simplex method
■■ Summary of Solution Methods that will solve both maximization and minimization problems with any combination
of … , Ú, and = problem constraints. The only requirement is that each problem
■■ Larger Problems: Refinery
Application
constraint must have a nonnegative constant on the right side. (This restriction is eas-
ily accommodated, as you will see.)
Introduction to the Big M Method

We introduce the big M method through a simple maximization problem with mixed
problem constraints. The key parts of the method will then be summarized and ap-
plied to more complex problems.
SECTION 5.4 Maximization and Minimization with Mixed Problem Constraints 309
Consider the following problem:
Maximize P = 2x1 + x2
(1)
- x1 + x2 Ú 2
x1, x2 Ú 0
To form an equation out of the first inequality, we introduce a slack variable s1, as
before, and write
x1 + x2 + s1 = 10
How can we form an equation out of the second inequality? We introduce a second
variable s2 and subtract it from the left side:
- x1 + x2 - s2 = 2
The variable s2 is called a surplus variable because it is the amount (surplus) by
which the left side of the inequality exceeds the right side.
Next, we express the linear programming problem (1) as a system of equations:
x1 + x2 + s1 = 10
- x1 + x2 - s2 = 2 (2)
- 2x1 - x2 + P = 0
x1, x2, s1, s2 Ú 0
It can be shown that a basic solution of system (2) is not feasible if any of the vari-
ables (excluding P) are negative. So a surplus variable is required to satisfy the
nonnegative constraint.
The basic solution found by setting the nonbasic variables x1 and x2 equal to 0 is
x1 = 0, x2 = 0, s1 = 10, s2 = - 2, P = 0
But this basic solution is not feasible, since the surplus variable s2 is negative (which
is a violation of the nonnegative requirements of all variables except P). The simplex
method works only when the basic solution for a tableau is feasible, so we cannot
solve this problem simply by writing the tableau for (2) and starting pivot operations.
To use the simplex method on problems with mixed constraints, we turn to an
ingenious device called an artificial variable. This variable has no physical meaning
in the original problem. It is introduced solely for the purpose of obtaining a basic
feasible solution so that we can apply the simplex method. An artificial variable is
a variable introduced into each equation that has a surplus variable. As before, to en-
sure that we consider only basic feasible solutions, an artificial variable is required
to satisfy the nonnegative constraint. (As we will see later, artificial variables are
also used to augment equality problem constraints when they are present.)
Returning to the problem at hand, we introduce an artificial variable a1 into the
equation involving the surplus variable s2:
- x1 + x2 - s2 + a1 = 2
To prevent an artificial variable from becoming part of an optimal solution to
the original problem, a very large “penalty” is introduced into the objective function.
This penalty is created by choosing a positive constant M so large that the artificial
variable is forced to be 0 in any final optimal solution of the original problem. (Since
the constant M can be as large as we wish in computer solutions, M is often selected
as the largest number the computer can hold!) Then we add the term - Ma1 to the
objective function:
P = 2x1 + x2 - Ma1
We now have a new problem, which we call the modified problem:
Maximize P = 2x1 + x2 - Ma1

subject to x1 + x2 + s1 = 10
(3)
- x1 + x2 - s2 + a1 = 2
x1, x2, s1, s2, a1 Ú 0
The initial system for the modified problem (3) is

x1 + x2 + s1 = 10
- x1 + x2 - s2 + a1 = 2
- 2x1 - x2 + Ma1 + P = 0 (4)
x1, x2, s1, s2, a1 Ú 0
We next write the augmented matrix for system (4), which we call the preliminary
simplex tableau for the modified problem. (The reason we call it the “preliminary”
simplex tableau instead of the “initial” simplex tableau will be made clear shortly.)
x1 x2 s1 s2 a1 P
1 1 1 0 0 0 10
(5)
C -1 1 0 -1 1 0 3 2S
-2 -1 0 0 M 1 0
To start the simplex process, including any necessary pivot operations, the pre-
liminary simplex tableau should either meet the two requirements given in the fol-
lowing box or be transformed by row operations into a tableau that meets these two
requirements.
DEFINITION Initial Simplex Tableau

For a system tableau to be considered an initial simplex tableau, it must satisfy the
following two requirements:
1. The requisite number of basic variables must be selectable by the process
described in Section 5.2. That is, a variable can be selected as a basic variable
only if it corresponds to a column in the tableau that has exactly one nonzero
element and the nonzero element in the column is not in the same row as the
nonzero element in the column of another basic variable. The remaining vari-
ables are then selected as nonbasic variables to be set equal to 0 in determining
a basic solution.
2. The basic solution found by setting the nonbasic variables equal to 0 is feasible.
Tableau (5) satisfies the first initial simplex tableau requirement, since s1, s2,
and P can be selected as basic variables according to the first requirement. (Not all
preliminary simplex tableaux satisfy the first requirement; see Example 2.) However,
tableau (5) does not satisfy the second initial simplex tableau requirement since the
basic solution is not feasible 1s2 = - 22. To use the simplex method, we must use
row operations to transform tableau (5) into an equivalent matrix that satisfies both
initial simplex tableau requirements. Note that this transformation is not a pivot
operation.
To get an idea of how to proceed, notice in tableau (5) that - 1 in the s2 column
is in the same row as 1 in the a1 column. This is not an accident! The artificial vari-
able a1 was introduced so that this would happen. If we eliminate M from the bot-
tom of the a1 column, the nonbasic variable a1 will become a basic variable and the
troublesome basic variable s2 will become a nonbasic variable. We proceed to elimi-

nate M from the a1 column using row operations:
x1 x2 s1 s2 a1 P
1 1 1 0 0 0 10
C -1 1 0 -1 1 0 3 2S
-2 -1 0 0 M 1 0 1 - M2R2 + R3 S R3
1 1 1 0 0 0 10
∼ C -1 1 0 -1 1 0 3 2S
M - 2 -M - 1 0 M 0 1 - 2M
From this last matrix we see that the basic variables are s1, a1, and P. The basic solu-
tion found by setting the nonbasic variables x1, x2, and s2 equal to 0 is
x1 = 0, x2 = 0, s1 = 10, s2 = 0, a1 = 2, P = - 2M
The basic solution is feasible (P can be negative), and both requirements for an initial
simplex tableau are met. We perform pivot operations to find an optimal solution.
The pivot column is determined by the most negative indicator in the bottom
row of the tableau. Since M is a positive number, - M - 1 is certainly a negative
indicator. What about the indicator M - 2? Remember that M is a very large positive
number. We will assume that M is so large that any expression of the form M - k is
positive. So the only negative indicator in the bottom row is - M - 1.
x1 x2 s1 s2 a1 P
10
1 1 1 0 0 0 10 1 = 10
Pivot row S C - 1 1 0 -1 1 03 2S 2
1 = 2
M - 2 -M - 1 0 M 0 1 - 2M
c
Pivot column
Having identified the pivot element, we now begin pivoting:
x1 x2 s1 s2 a1 P
s1 1 1 1 0 0 0 10 1 - 12R2 + R1 S R1
a1 C - 1 1 0 -1 1 0 3 2S
P M - 2 -M - 1 0 M 0 1 - 2M 1M + 12R2 + R3 S R3
1
s1 2 0 1 1 -1 0 8 2 R1 S R1
∼ 2C
x -1 1 0 -1 1 0 3 2S
P -3 0 0 -1 M + 1 1 2
1 1
1 0 2 2 - 12 0 4
∼C -1 1 0 -1 1 0 3 2S R1 + R2 S R2
-3 0 0 -1 M + 1 1 2 3R1 + R3 S R3
1 1 1
x1 1 0 2 2 2 0 4
∼ x2 C 0 1 1
2 - 12 1
2 0 3 6S
3 1 1
P 0 0 2 2 M - 2 1 14
Since all the indicators in the last row are nonnegative (M - 12 is nonnegative be-
cause M is a very large positive number), we can stop and write the optimal solution:
Max P = 14 at x1 = 4, x2 = 6, s1 = 0, s2 = 0, a1 = 0
This is an optimal solution to the modified problem (3). How is it related to the origi-
nal problem (2)? Since a1 = 0 in this solution,
x1 = 4, x2 = 6, s1 = 0, s2 = 0, P = 14 (6)
is certainly a feasible solution for system (2). [You can verify this by direct substitu-
tion into system (2).] Surprisingly, it turns out that solution (6) is an optimal solution
to the original problem. To see that this is true, suppose we were able to find feasible
values of x1, x2, s1, and s2 that satisfy the original system (2) and produce a value of
P 7 14. Then by using these same values in problem (3) along with a1 = 0, we
would have a feasible solution of problem (3) with P 7 14. This contradicts the fact
that P = 14 is the maximum value of P for the modified problem. Solution (6) is an
optimal solution for the original problem.
As this example illustrates, if a1 = 0 is an optimal solution for the modified prob-
lem, then deleting a1 produces an optimal solution for the original problem. What hap-
pens if a1 ∙ 0 in the optimal solution for the modified problem? In this case, it can be
shown that the original problem has no optimal solution because its feasible set is empty.
In larger problems, each Ú problem constraint will require the introduction
of a surplus variable and an artificial variable. If one of the problem constraints is
an equation rather than an inequality, then there is no need to introduce a slack or
surplus variable. However, each = problem constraint will require the introduction
of another artificial variable to prevent the initial basic solution from violating the
equality constraint—the decision variables are often 0 in the initial basic solution
(see Example 2). Finally, each artificial variable also must be included in the objec-
tive function for the modified problem. The same constant M can be used for each
artificial variable. Because of the role that the constant M plays in this approach, this
method is often called the big M method.
Big M Method
We summarize the key steps of the big M method and use them to solve several
problems.
PROCEDURE Big M Method: Introducing Slack, Surplus, and Artificial

Variables to Form the Modified Problem
Step 1 If any problem constraints have negative constants on the right side, mul-
tiply both sides by - 1 to obtain a constraint with a nonnegative constant.
(If the constraint is an inequality, this will reverse the direction of the
inequality.)
Step 2 Introduce a slack variable in each … constraint.
Step 3 Introduce a surplus variable and an artificial variable in each Ú constraint.
Step 4 Introduce an artificial variable in each = constraint.
Step 5 For each artificial variable ai, add -Mai to the objective function. Use the
same constant M for all artificial variables.
EXAMPLE 1 Finding the Modified Problem Find the modified problem for the following linear
programming problem. (Do not attempt to solve the problem.)
Maximize P = 2x1 + 5x2 + 3x3
subject to x1 + 2x2 - x3 … 7
- x1 + x2 - 2x3 … -5
x1 + 4x2 + 3x3 Ú 1
2x1 - x2 + 4x3 = 6
x1, x2, x3 Ú 0
SOLUTION First, we multiply the second constraint by - 1 to change - 5 to 5:
1 - 12 1 - x1 + x2 - 2x3 2 Ú 1 - 12 1 - 52
x1 - x2 + 2x3 Ú 5
Next, we introduce the slack, surplus, and artificial variables according to the pro-
cedure stated in the box:
x1 + 2x2 - x3 + s1 = 7
x1 - x2 + 2x3 - s2 + a1 = 5
x1 + 4x2 + 3x3 - s3 + a2 = 1
2x1 - x2 + 4x3 + a3 = 6
Finally, we add - Ma1, -Ma2, and -Ma3 to the objective function:
P = 2x1 + 5x2 + 3x3 - Ma1 - Ma2 - Ma3
The modified problem is
Maximize P = 2x1 + 5x2 + 3x3 - Ma1 - Ma2 - Ma3
subject to x1 + 2x2 - x3 + s1 = 7
x1 - x2 + 2x3 - s2 + a1 = 5
x1 + 4x2 + 3x3 - s3 + a2 = 1
2x1 - x2 + 4x3 + a3 = 6
x1, x2, x3, s1, s2, s3, a1, a2, a3 Ú 0
Matched Problem 1 Repeat Example 1 for

Maximize P = 3x1 - 2x2 + x3
subject to x1 - 2x2 + x3 Ú 5
- x1 - 3x2 + 4x3 … - 10
2x1 + 4x2 + 5x3 … 20
3x1 - x2 - x3 = - 15
x1, x2, x3 Ú 0
Now we can list the key steps for solving a problem using the big M method. The
various steps and remarks are based on a number of important theorems, which we
assume without proof. In particular, step 2 is based on the fact that (except for some
degenerate cases not considered here) if the modified linear programming problem
has an optimal solution, then the preliminary simplex tableau will be transformed
into an initial simplex tableau by eliminating the M’s from the columns correspond-
ing to the artificial variables in the preliminary simplex tableau. Having obtained an
initial simplex tableau, we perform pivot operations.
PROCEDURE Big M Method: Solving the Problem

Step 1 Form the preliminary simplex tableau for the modified problem.
Step 2 Use row operations to eliminate the M’s in the bottom row of the preliminary
simplex tableau in the columns corresponding to the artificial variables. The
resulting tableau is the initial simplex tableau.
Step 3 Solve the modified problem by applying the simplex method to the initial
simplex tableau found in step 2.
Step 4 Relate the optimal solution of the modified problem to the original problem.
(A) If the modified problem has no optimal solution, then the original prob-
lem has no optimal solution.
(B) If all artificial variables are 0 in the optimal solution to the modified
problem, then delete the artificial variables to find an optimal solution
to the original problem.
(C) If any artificial variables are nonzero in the optimal solution to the
modified problem, then the original problem has no optimal solution.
EXAMPLE 2 Using the Big M Method Solve the following linear programming problem using
the big M method:
Maximize P = x1 - x2 + 3x3
x1 + x3 = 5
x2 + x3 Ú 10
x1, x2, x3 Ú 0
SOLUTION State the modified problem:
Maximize P = x1 - x2 + 3x3 - Ma1 - Ma2

subject to x1 + x2 + s1 = 20
x1 + x3 + a1 = 5
x2 + x3 - s2 + a2 = 10
x1, x2, x3, s1, s2, a1, a2 Ú 0
Write the preliminary simplex tableau for the modified problem, and find the initial
simplex tableau by eliminating the M’s from the artificial variable columns:
x1 x2 x3 s1 a1 s2 a2 P
1 1 0 1 0 0 0 0 20 Eliminate M from the a1
1 0 1 0 1 0 0 0 4 5 column
≥ ¥
0 1 1 0 0 -1 1 0 10
-1 1 -3 0 M 0 M 1 0 1 - M2R2 + R4 S R4
1 1 0 1 0 0 0 0 20 Eliminate M from the a2

1 0 1 0 1 0 0 0 5 column
∼≥ 4 ¥
0 1 1 0 0 -1 1 0 10
-M - 1 1 -M - 3 0 0 0 M 1 - 5M 1 - M2R3 + R4 S R4
1 1 0 1 0 0 0 0 20
1 0 1 0 1 0 0 0 4 5
∼≥ ¥
0 1 1 0 0 -1 1 0 10
-M - 1 -M + 1 - 2M - 3 0 0 M 0 1 - 15M
From this last matrix we see that the basic variables are s1, a1, a2, and P. The basic
solution found by setting the nonbasic variables x1, x2, x3, and s2 equal to 0 is
x1 = 0, x2 = 0, x3 = 0, s1 = 20, a1 = 5, s2 = 0, a2 = 10, P = - 15M
The basic solution is feasible, and both requirements for an initial simplex tableau
are met. We perform pivot operations to find the optimal solution.
x1 x2 x3 s1 a1 s2 a2 P
s1 1 1 0 1 0 0 0 0 20
a1 1 0 1 0 1 0 0 04 5
≥ ¥
a2 0 1 1 0 0 -1 1 0 10 1 - 12R2 + R3 S R3
P -M - 1 -M + 1 - 2M - 3 0 0 M 0 1 - 15M 12M + 32R2 + R4 S R4
s1 1 1 0 1 0 0 0 0 20 1 - 12R3 + R1 S R1
x3 1 0 1 0 1 0 0 04 5
∼ ≥ ¥
a2 -1 1 0 0 -1 -1 1 0 5
P M + 2 -M + 1 0 0 2M + 3 M 0 1 - 5M + 15 1M - 12R3 + R4 S R4
s1 2 0 0 1 1 1 -1 0 15
x3 1 0 1 0 1 0 0 04 5
∼ ≥ ¥
x2 -1 1 0 0 -1 -1 1 0 5
P 3 0 0 0 M + 4 1 M - 1 1 10
Since the bottom row has no negative indicators, we can stop and write the
optimal solution to the modified problem:
x1 = 0, x2 = 5, x3 = 5, s1 = 15, a1 = 0, s2 = 0, a2 = 0, P = 10
Since a1 = 0 and a2 = 0, the solution to the original problem is
Max P = 10 at x1 = 0, x2 = 5, x3 = 5

the big M method:
Maximize P = x1 + 4x2 + 2x3
x1 - x3 = 6
x1 - x2 - x3 Ú 1
x1, x2, x3 Ú 0
EXAMPLE 3 Using the Big M Method Solve the following linear programming problem using
the big M method:
x1 + 2x2 Ú 10
x1, x2 Ú 0
SOLUTION Introducing slack, surplus, and artificial variables, we obtain the modi-
fied problem:
2x1 + x2 + s1 = 4
x1 + 2x2 - s2 + a1 = 10 Modified problem
- 3x1 - 5x2 + Ma1 + P = 0
Preliminary simplex tableau

x1 x2 s1 s2 a1 P
2 1 1 0 0 0 4 Eliminate M in the a1 column.
C 1 2 0 -1 1 0 3 10 S
-3 -5 0 0 M 1 0 1 - M2R2 + R3 S R3
Initial simplex tableau
s1 2 1 1 0 0 0 4 Begin pivot operations.
∼ a1 C 1 2 0 -1 1 0 3 10 S 1 - 22R1 + R2 S R2
P -M - 3 - 2M - 5 0 M 0 1 - 10M 12M + 52R1 + R3 S R3
x2 2 1 1 0 0 0 4
∼ a1 C -3 0 -2 -1 1 0 3 2 S
P 3M + 7 0 2M + 5 M 0 1 - 2M + 20
The optimal solution of the modified problem is

x1 = 0, x2 = 4, s1 = 0, s2 = 0,
a1 = 2, P = - 2M + 20
Since a1 ∙ 0, the original problem has no optimal solution. Figure 1 shows that the
feasible region for the original problem is empty.
x2
A
5 x1 1 2x2 $ 10
x1 $ 0, x2 $ 0
B
x1
0 5 10
2x1 1 x2 # 4
x1 $ 0, x2 $ 0
Figure 1 A>B5[

the big M method:
2x1 + x2 Ú 12
x1, x2 Ú 0
y
Minimization by the Big M Method
y 5 f (x)
In addition to solving any maximization problem, the big M method can be used
to solve minimization problems. To minimize an objective function, we need only
m
to maximize its negative. Figure 2 illustrates the fact that the minimum value of
x a function f occurs at the same point as the maximum value of the function - f .
2m Furthermore, if m is the minimum value of f, then - m is the maximum value of - f ,
and conversely. So we can find the minimum value of a function f by finding the
y 5 2f (x) maximum value of - f and then changing the sign of the maximum value.
Figure 2
EXAMPLE 4 Production Scheduling: Minimization Problem A small jewelry manufacturer

hires a highly skilled gem cutter to work at least 6 hours per day. On the other hand,
the polishing facilities can be used for at most 10 hours per day. The company
specializes in three kinds of semiprecious gemstones, J, K, and L. Relevant cutting,
polishing, and cost requirements are listed in the table. How many gemstones of
each type should be processed each day to minimize the cost of the finished stones?
What is the minimum cost?
J K L
Cutting 1 hr 1 hr 1 hr
Polishing 2 hr 1 hr 2 hr
Cost per stone $30 $30 $10
SOLUTION Since we must decide how many gemstones of each type should be
processed each day, the decision variables are
x1 = number of type J gemstones processed each day
x2 = number of type K gemstones processed each day
x3 = number of type L gemstones processed each day
Since the data is already summarized in a table, we can proceed directly to the model:
Minimize C = 30x1 + 30x2 + 10x3 Objective function
subject to x1 + x2 + x3 Ú 6
f Problem constraints
2x1 + x2 + 2x3 … 10
x1, x2, x3 Ú 0 Nonnegative constraints
We convert this to a maximization problem by letting
P = - C = - 30x1 - 30x2 - 10x3
We get
Maximize P = - 30x1 - 30x2 - 10x3
subject to x1 + x2 + x3 Ú 6
2x1 + x2 + 2x3 … 10
x1, x2, x3 Ú 0
and Min C = - Max P. To solve, we first state the modified problem:
x1 + x2 + x3 - s1 + a1 = 6
2x1 + x2 + 2x3 + s2 = 10
30x1 + 30x2 + 10x3 + Ma1 + P = 0
x1, x2, x3, s1, s2, a1 Ú 0
x1 x2 x3 s1 a1 s2 P
1 1 1 -1 1 0 0 6 Eliminate M in the a1 column
C 2 1 2 0 0 1 0 † 10 S 1 - M2R1 + R3 S R3
30 30 10 0 M 0 1 0
Begin pivot operations. Assume M is so large
that - M + 30 and - M + 10 are negative
a1 1 1 1 -1 1 0 0 6
s
∼ 2 C 2 1 2 0 0 1 0 † 10 S 0.5R2 S R2
P - M + 30 - M + 30 - M + 10 M 0 0 1 - 6M
1 1 1 -1 1 0 0 6 1 - 12R2 + R1 S R1
∼C 1 0.5 1 0 0 0.5 0 † 5 S
- M + 30 - M + 30 - M + 10 M 0 0 1 - 6M 1M - 102R2 + R3 S R3
a1 0 0.5 0 -1 1 - 0.5 0 1 2R1 S R1

∼ x3 C 1 0.5 1 0 0 0.5 0† 5 S
P 20 - 0.5M + 25 0 M 0 0.5M - 5 1 - M - 50
0 1 0 -2 2 -1 0 2 1 - 0.52R1 + R2 S R2
∼C 1 0.5 1 0 0 0.5 0 † 5 S 10.5M - 252R1 + R3 S R3
20 - 0.5M + 25 0 M 0 0.5M - 5 1 - M - 50
x2 0 1 0 -2 2 -1 0 2
∼ x3 C 1 0 1 1 -1 1 0 † 4 S
P 20 0 0 50 M - 50 20 1 - 100
The bottom row has no negative indicators, so the optimal solution for the mod-
ified problem is
x1 = 0, x2 = 2, x3 = 4, s1 = 0, a1 = 0, s2 = 0, P = - 100
Since a1 = 0, deleting a1 produces the optimal solution to the original maximiza-
tion problem and also to the minimization problem. Thus,
Min C = - Max P = - 1 - 1002 = 100 at x1 = 0, x2 = 2, x3 = 4
That is, a minimum cost of $100 for gemstones will be realized if no type J, 2 type
K, and 4 type L stones are processed each day.
Matched Problem 4 Repeat Example 4 if the gem cutter works at least 8

hours a day and all other data remain the same.
Summary of Solution Methods

The big M method can be used to solve any minimization problems, including those
that can be solved by the dual method. (Note that Example 4 could have been solved
by the dual method.) Both methods of solving minimization problems are important.
You will be instructed to solve most minimization problems in Exercise 5.4 by the
big M method in order to gain more experience with this method. However, if the
method of solution is not specified, the dual method is usually easier.
Table 1 should help you select the proper method of solution for any linear pro-
gramming problem.
Table 1 Summary of Problem Types and Simplex Solution Methods

Problem Right-Side Coefficients of
Problem Type Constraints Constants Objective Function Method of Solution
1. Maximization … Nonnegative Any real numbers Simplex method with slack variables
2. Minimization Ú Any real numbers Nonnegative Form dual problem and solve by the preceding method
3. Maximization Mixed Nonnegative Any real numbers Form modified problem with slack, surplus, and
1 …, Ú, = 2 artificial variables, and solve by the big M method
4. Minimization Mixed Nonnegative Any real numbers Maximize negative of objective function by the preceding
1 …, Ú, = 2 method
Larger Problems: Refinery Application

Up to this point, all the problems we have considered could be solved by hand.
However, the real value of the simplex method lies in its ability to solve problems
with a large number of variables and constraints, where a computer is generally used
to perform the actual pivot operations. As a final application, we consider a problem
that would require the use of a computer to complete the solution.
EXAMPLE 5 Petroleum Blending A refinery produces two grades of gasoline, regular and pre-
mium, by blending together two components, A and B. Component A has an octane
rating of 90 and costs $28 a barrel. Component B has an octane rating of 110 and
costs $32 a barrel. The octane rating for regular gasoline must be at least 95, and the
octane rating for premium must be at least 105. Regular gasoline sells for $34 a bar-
rel and premium sells for $40 a barrel. Currently, the company has 30,000 barrels
of component A and 20,000 barrels of component B. It also has orders for 20,000
barrels of regular and 10,000 barrels of premium that must be filled. Assuming that
all the gasoline produced can be sold, determine the maximum possible profit.
SOLUTION This problem is similar to the transportation problem in Section 5.3.
That is, to maximize the profit, we must decide how much of each component must
be used to produce each grade of gasoline. Thus, the decision variables are
x1 = number of barrels of component A used in regular gasoline
x2 = number of barrels of component A used in premium gasoline
x3 = number of barrels of component B used in regular gasoline
x4 = number of barrels of component B used in premium gasoline
Table 2
Component Octane Rating Cost ($) Available Supply
A 90 28 30,000 barrels
B 110 32 20,000 barrels
Grade Minimum Octane Rating Selling Price($) Existing Orders
Regular 95 34 20,000 barrels
Premium 105 40 10,000 barrels
Next, we summarize the data in table form (Table 2). Once again, we have to adjust
the form of the table to fit the data.
The total amount of component A used is x1 + x2. This cannot exceed the avail-
able supply. Thus, one constraint is
x1 + x2 … 30,000
The corresponding inequality for component B is
x3 + x4 … 20,000
The amounts of regular and premium gasoline produced must be sufficient to meet
the existing orders:
x1 + x3 Ú 20,000 Regular
x2 + x4 Ú 10,000 Premium
Now consider the octane ratings. The octane rating of a blend is simply the
proportional average of the octane ratings of the components. So the octane rating
for regular gasoline is
x1 x3
90 + 110
x1 + x3 x1 + x3
where x1 > 1x1 + x3 2 is the percentage of component A used in regular gasoline and
x3 > 1x1 + x3 2 is the percentage of component B. The final octane rating of regular
gasoline must be at least 95; so
x1 x3
90 + 110 Ú 95 Multiply by x1 + x3.
x1 + x3 x1 + x3
90x1 + 110x3 Ú 951x1 + x3 2 Collect like terms on the right side.
0 Ú 5x1 - 15x3 Octane rating for regular
The corresponding inequality for premium gasoline is
x2 x4
90 + 110 Ú 105
x2 + x4 x2 + x4
90x2 + 110x4 Ú 1051x2 + x4 2
0 Ú 15x2 - 5x4 Octane rating for premium
The cost of the components used is
C = 281x1 + x2 2 + 321x3 + x4 2
The revenue from selling all the gasoline is
R = 341x1 + x3 2 + 401x2 + x4 2
and the profit is
P = R - C
= 341x1 + x3 2 + 401x2 + x4 2 - 281x1 + x2 2 - 321x3 + x4 2
= 134 - 282x1 + 140 - 282x2 + 134 - 322x3 + 140 - 322x4
= 6x1 + 12x2 + 2x3 + 8x4
To find the maximum profit, we must solve the following linear programming problem:
Maximize P = 6x1 + 12x2 + 2x3 + 8x4 Profit
subject to x1 + x2 … 30,000 Available A
x3 + x4 … 20,000 Available B
x1 + x3 Ú 20,000 Required regular
x2 + x4 Ú 10,000 Required premium
5x1 - 15x3 … 0 Octane for regular
15x2 - 5x4 … 0 Octane for premium
x1, x2, x3, x4 Ú 0
We will use technology to solve this large problem. There are many types of soft-
ware that use the big M method to solve linear programming problems, including Java
applets, graphing calculator programs, and spreadsheets. Because you are likely to use
different software than we did, we will simply display the initial and final tableaux.
Notice that in the last row of the initial tableau, we entered a large number, 106, instead
of the symbol M. This is typical of software implementations of the big M method.
x1 x2 x3 x4 s1 s2 s3 a1 s4 a2 s5 s6 P
1 1 0 0 1 0 0 0 0 0 0 0 0 30,000
0 0 1 1 0 1 0 0 0 0 0 0 0 20,000
1 0 1 0 0 0 -1 1 0 0 0 0 0 20,000
G 0 1 0 1 0 0 0 0 -1 1 0 0 0 7 10,000 W
5 0 - 15 0 0 0 0 0 0 0 1 0 0 0
0 15 0 -5 0 0 0 0 0 0 0 1 0 0
-6 - 12 -2 -8 0 0 0 106 0 10 6
0 0 1 0
The final table produced by the software is

x1 x2 x3 x4 s1 s2 s3 a1 s4 a2 s5 s6 P
0 0 0 0 1.5 - 0.5 1 -1 0 0 - 0.1 - 0.1 0 15,000
0 0 0 0 - 0.5 1.5 0 0 1 -1 0.1 0.1 0 5,000
0 0 1 0 0.375 - 0.125 0 0 0 0 - 0.075 - 0.025 0 8,750
G0 0 0 1 - 0.375 1.125 0 0 0 0 0.075 0.025 7
0 11,250 W
1 0 0 0 1.125 - 0.375 0 0 0 0 - 0.025 - 0.075 0 26,250
0 1 0 0 - 0.125 0.375 0 0 0 0 0.025 0.075 0 3,750
0 0 0 0 3 11 0 106 0 106 0.6 0.6 1 310,000
From the final tableau, we see that the refinery should blend 26,250 barrels of com-
ponent A and 8,750 barrels of component B to produce 35,000 barrels of regular
gasoline. They should blend 3,750 barrels of component A and 11,250 barrels of
component B to produce 15,000 barrels of premium gasoline. This will result in a
maximum profit of $310,000.

Interpret the values of the slack and surplus variables in the computer solution to
Example 5.
Matched Problem 5 Suppose that the refinery in Example 5 has 35,000 bar-
rels of component A, which costs $25 a barrel, and 15,000 barrels of component
B, which costs $35 a barrel. If all other data remain the same, formulate a linear
programming problem to find the maximum profit. Do not attempt to solve the
problem (unless you have access to software that can solve linear programming
problems).
Exercises 5.4
A In Problems 1–8, 3. Maximize P = 3x1 + 5x2
(A) Introduce slack, surplus, and artificial variables and form the subject to 2x1 + x2 … 8
modified problem. x1 + x2 = 6
(B) Write the preliminary simplex tableau for the modified prob- x1, x2 Ú 0
lem and find the initial simplex tableau. 4. Maximize P = 4x1 + 3x2
(C) Find the optimal solution of the modified problem by apply- subject to x1 + 3x2 … 24
ing the simplex method to the initial simplex tableau. x1 + x2 = 12
(D) Find the optimal solution of the original problem, if it exists. x1, x2 Ú 0
1. Maximize P = 5x1 + 2x2 5. Maximize P = 4x1 + 3x2

subject to x1 + 2x2 … 12 subject to - x1 + 2x2 … 2
x1 + x2 Ú 4 x1 + x2 Ú 4
x1, x2 Ú 0 x1, x2 Ú 0
2. Maximize P = 3x1 + 7x2 6. Maximize P = 3x1 + 4x2

subject to 2x1 + x2 … 16 subject to x1 - 2x2 … 2
x1 + x2 Ú 6 x1 + x2 Ú 5
x1, x2 Ú 0 x1, x2 Ú 0
7. Maximize P = 5x1 + 10x2 17. Maximize P = 3x1 + 5x2 + 6x3

subject to x1 + x2 … 3 subject to 2x1 + x2 + 2x3 … 8
2x1 + 3x2 Ú 12 2x1 + x2 - 2x3 = 0
x1, x2 Ú 0 x1, x2, x3 Ú 0
8. Maximize P = 4x1 + 6x2 18. Maximize P = 3x1 + 6x2 + 2x3
subject to x1 + x2 … 2 subject to 2x1 + 2x2 + 3x3 … 12
3x1 + 5x2 Ú 15 2x1 - 2x2 + x3 = 0
x1, x2 Ú 0 x1, x2, x3 Ú 0
B Use the big M method to solve Problems 9–22. 19. Maximize P = 2x1 + 3x2 + 4x3
subject to x1 + 2x2 + x3 … 25
9. Minimize and maximize P = 2x1 - x2
2x1 + x2 + 2x3 … 60
x1 + 2x2 - x3 Ú 10
5x1 + 3x2 Ú 30
x1, x2, x3 Ú 0
x1, x2 Ú 0
20. Maximize P = 5x1 + 2x2 + 9x3
10. Minimize and maximize P = - 4x1 + 16x2
subject to 2x1 + 4x2 + x3 … 150
3x1 + 3x2 + x3 … 90
x1 + 2x2 Ú 16
- x1 + 5x2 + x3 Ú 120
x1, x2 Ú 0
x1, x2, x3 Ú 0
21. Maximize P = x1 + 2x2 + 5x3
subject to x1 + 3x2 + 2x3 … 60
2x1 + x2 … 21
2x1 + 5x2 + 2x3 Ú 50
x1 + x2 Ú 10
x1 - 2x2 + x3 Ú 40
x1, x2 Ú 0
x1, x2, x3 Ú 0
22. Maximize P = 2x1 + 4x2 + x3
subject to 2x1 + 3x2 + 5x3 … 280
2x1 + x2 … 16
2x1 + 2x2 + x3 Ú 140
x1 + x2 Ú 9
2x1 + x2 Ú 150
x1, x2 Ú 0
x1, x2, x3 Ú 0
13. Maximize P = 10x1 + 12x2 + 20x3
23. Solve Problems 5 and 7 by graphing (the geometric method).
x1 - x2 + 2x3 = 6 24. Solve Problems 6 and 8 by graphing (the geometric method).
x1, x2, x3 Ú 0 C Problems 25–32 are mixed. Some can be solved by the methods
14. Maximize P = 5x1 + 7x2 + 9x3 presented in Sections 5.2 and 5.3 while others must be solved by
the big M method.
subject to x1 - x2 + x3 Ú 20
2x1 + x2 + 5x3 = 35 25. Minimize C = 10x1 - 40x2 - 5x3
x1, x2, x3 Ú 0 subject to x1 + 3x2 … 6
4x2 + x3 … 3
15. Minimize C = - 5x1 - 12x2 + 16x3
x1, x2, x3 Ú 0
subject to x1 + 2x2 + x3 … 10
2x1 + 3x2 + x3 Ú 6 26. Maximize P = 7x1 - 5x2 + 2x3
2x1 + x2 - x3 = 1 subject to x1 - 2x2 + x3 Ú - 8
x1, x2, x3 Ú 0 x1 - x2 + x3 … 10
x1, x2, x3 Ú 0
16. Minimize C = - 3x1 + 15x2 - 4x3
subject to 2x1 + x2 + 3x3 … 24 27. Maximize P = - 5x1 + 10x2 + 15x3
x1 + 2x2 + x3 Ú 6 subject to 2x1 + 3x2 + x3 … 24
x1 - 3x2 + x3 = 2 x1 - 2x2 - 2x3 Ú 1
x1, x2, x3 Ú 0 x1, x2, x3 Ú 0
28. Minimize C = - 5x1 + 10x2 + 15x3 Units per Bottle

subject to 2x1 + 3x2 + x3 … 24 A B C
x1 - 2x2 - 2x3 Ú 1 Protein 10 10 20
x1, x2, x3 Ú 0 Carbohydrates 2 3 4
Cost per bottle ($) 0.60 0.40 0.90
29. Minimize C = 10x1 + 40x2 + 5x3
subject to x1 + 3x2 Ú 6 36. Human nutrition. Discuss the effect on the solution to
4x2 + x3 Ú 3 Problem 35 if the cost of brand C liquid diet food increases to
x1, x2, x3 Ú 0 $1.50 per bottle.
30. Maximize P = 8x1 + 2x2 - 10x3 37. Plant food. A farmer can use three types of plant food: mix
subject to x1 + x2 - 3x3 … 6 A, mix B, and mix C. The amounts (in pounds) of nitrogen,
phosphoric acid, and potash in a cubic yard of each mix are
4x1 - x2 + 2x3 … - 7 given in the table. Tests performed on the soil indicate that
x1, x2, x3 Ú 0 the field needs at least 800 pounds of potash. The tests also
indicate that no more than 700 pounds of phosphoric acid
31. Maximize P = 12x1 + 9x2 + 5x3
should be added to the field. The farmer plans to plant a crop
subject to x1 + 3x2 + x3 … 40 that requires a great deal of nitrogen. How many cubic yards
2x1 + x2 + 3x3 … 60 of each mix should be added to the field in order to satisfy
x1, x2, x3 Ú 0 the potash and phosphoric acid requirements and maximize
the amount of nitrogen? What is the maximum amount of
32. Minimize C = 10x1 + 12x2 + 28x3 nitrogen?
subject to 4x1 + 2x2 + 3x3 Ú 20
Pounds per Cubic Yard
3x1 - x2 - 4x3 … 10
A B C
x1, x2, x3 Ú 0 Nitrogen 12 16 8
Phosphoric acid 12 8 16
Potash 16 8 16
Applications 38. Plant food. Discuss the effect on the solution to Problem 37
In Problems 33–38, construct a mathematical model in the form if the limit on phosphoric acid is increased to 1,000 pounds.
of a linear programming problem. (The answers in the back of In Problems 39–47, construct a mathematical model in the form
the book for these application problems include the model.) Then of a linear programming problem. Do not solve.
solve the problem using the big M method.
39. Manufacturing. A company manufactures car and truck
33. Advertising. An advertising company wants to attract new frames at plants in Milwaukee and Racine. The Milwau-
customers by placing a total of at most 10 ads in 3 newspa- kee plant has a daily operating budget of $50,000 and can
pers. Each ad in the Sentinel costs $200 and will be read by produce at most 300 frames daily in any combination. It
2,000 people. Each ad in the Journal costs $200 and will be costs $150 to manufacture a car frame and $200 to manufac-
read by 500 people. Each ad in the Tribune costs $100 and ture a truck frame at the Milwaukee plant. The Racine plant
will be read by 1,500 people. The company wants at least has a daily operating budget of $35,000, and can produce a
16,000 people to read its ads. How many ads should it place maximum combined total of 200 frames daily. It costs $135
in each paper in order to minimize the advertising costs? to manufacture a car frame and $180 to manufacture a truck
What is the minimum cost? frame at the Racine plant. Based on past demand, the company
34. Advertising. Discuss the effect on the solution to Problem wants to limit production to a maximum of 250 car frames
33 if the Tribune will not accept more than 4 ads from the and 350 truck frames per day. If the company realizes a profit
company. of $50 on each car frame and $70 on each truck frame, how
many frames of each type should be produced at each plant to
35. Human nutrition. A person on a high-protein, low- maximize the daily profit?
carbohydrate diet requires at least 100 units of protein and at
most 24 units of carbohydrates daily. The diet will consist 40. Loan distributions. A savings and loan company has $3
entirely of three special liquid diet foods: A, B, and C. The million to lend. The types of loans and annual returns offered
contents and costs of the diet foods are given in the table. are given in the table. State laws require that at least 50% of
How many bottles of each brand of diet food should be the money loaned for mortgages must be for first mortgages
consumed daily in order to meet the protein and carbohydrate and that at least 30% of the total amount loaned must be for
requirements at minimal cost? What is the minimum cost? either first or second mortgages. Company policy requires
that the amount of signature and automobile loans cannot 44. Investment strategy. Refer to Problem 43. Suppose the
exceed 25% of the total amount loaned and that signature investor decides that she would like to minimize the total risk
loans cannot exceed 15% of the total amount loaned. How factor, as long as her return does not fall below 9%. What
much money should be allocated to each type of loan in order percentage of her total investments should be invested in each
to maximize the company’s return? choice to minimize the total risk level?
Type of Loan Annual Return (%) 45. Human nutrition. A dietitian arranges a special diet using
Signature 18 foods L, M, and N. The table gives the nutritional contents
and cost of 1 ounce of each food. The diet’s daily require-
First mortgage 12
ments are at least 400 units of calcium, at least 200 units of
Second mortgage 14 iron, at least 300 units of vitamin A, at most 150 units of
Automobile 16 cholesterol, and at most 900 calories. How many ounces of
each food should be used in order to meet the diet’s require-
41. Oil refining. A refinery produces two grades of gasoline, ments at a minimal cost?
regular and premium, by blending together three components:
A, B, and C. Component A has an octane rating of 90 and Units per Bottle
costs $28 a barrel, component B has an octane rating of 100 L M N
and costs $30 a barrel, and component C has an octane rating
Calcium 30 10 30
of 110 and costs $34 a barrel. The octane rating for regular
must be at least 95 and the octane rating for premium must be Iron 10 10 10
at least 105. Regular gasoline sells for $38 a barrel and pre- Vitamin A 10 30 20
mium sells for $46 a barrel. The company has 40,000 barrels Cholesterol 8 4 6
of component A, 25,000 barrels of component B, and 15,000 Calories 60 40 50
barrels of component C. It must produce at least 30,000 bar-
Cost per ounce ($) 0.40 0.60 0.80
rels of regular and 25,000 barrels of premium. How should
the components be blended in order to maximize profit?
46. Mixing feed. A farmer grows three crops: corn, oats, and
42. Trail mix. A company makes two brands of trail mix, regular soybeans. He mixes them to feed his cows and pigs. At least
and deluxe, by mixing dried fruits, nuts, and cereal. The recipes 40% of the feed mix for the cows must be corn. The feed mix
for the mixes are given in the table. The company has 1,200 for the pigs must contain at least twice as much soybeans as
pounds of dried fruits, 750 pounds of nuts, and 1,500 pounds of corn. He has harvested 1,000 bushels of corn, 500 bushels of
cereal for the mixes. The company makes a profit of $0.40 on oats, and 1,000 bushels of soybeans. He needs 1,000 bushels
each pound of regular mix and $0.60 on each pound of deluxe of each feed mix for his livestock. The unused corn, oats, and
mix. How many pounds of each ingredient should be used in soybeans can be sold for $4, $3.50, and $3.25 a bushel, re-
each mix in order to maximize the company’s profit? spectively (thus, these amounts also represent the cost of the
crops used to feed the livestock). How many bushels of each
Type of Mix Ingredients crop should be used in each feed mix in order to produce suf-
Regular At least 20% nuts ficient food for the livestock at a minimal cost?
At most 40% cereal
47. Transportation. Three towns are forming a consolidated
Deluxe At least 30% nuts school district with two high schools. Each high school has a
At most 25% cereal maximum capacity of 2,000 students. Town A has 500 high
school students, town B has 1,200, and town C has 1,800.
43. Investment strategy. An investor is planning to divide her The weekly costs of transporting a student from each town
investments among high-tech mutual funds, global mutual to each school are given in the table. In order to balance the
funds, corporate bonds, municipal bonds, and CDs. Each of enrollment, the school board decided that each high school
these investments has an estimated annual return and a risk fac- must enroll at least 40% of the total student population.
tor (see the table). The risk level for each choice is the product Furthermore, no more than 60% of the students in any town
of its risk factor and the percentage of the total funds invested should be sent to the same high school. How many students
in that choice. The total risk level is the sum of the risk levels from each town should be enrolled in each school in order to
for all the investments. The investor wants at least 20% of her meet these requirements and minimize the cost of transport-
investments to be in CDs and does not want the risk level to ing the students?
exceed 1.8. What percentage of her total investments should be
Weekly Transportation Cost per Student ($)
invested in each choice to maximize the return?
School I School II
Investment Annual Return (%) Risk Factor Town A 4 8
High-tech funds 11 2.7 Town B 6 4
Global funds 10 1.8 Town C 3 9
Corporate bonds 9 1.2
Muncipal bonds 8 0.5
CDs 5 0
Answers to Matched Problems 4. A minimum cost of $200 is realized when no type J,

6 type K, and 2 type L stones are processed each day.
1. Maximize P = 3x1 - 2x2 + x3 - Ma1 - Ma2 - Ma3 5. Maximize P = 9x1 + 15x2 - x3 + 5x4
subject to x1 - 2x2 + x3 - s1 + a1 = 5 subject to x1 + x2 … 35,000
x1 + 3x2 - 4x3 - s2 + a2 = 10 x3 + x4 … 15,000
2x1 + 4x2 + 5x3 + s3 = 20 x1 + x3 Ú 20,000
- 3x1 + x2 + x3 + a3 = 15 x2 + x4 Ú 10,000
x1, x2, x3, s1, a1, s2, a2, s3, a3 Ú 0 5x1 - 15x3 … 0
2. Max P = 22 at x1 = 6, x2 = 4, x3 = 0 15x2 - 5x4 … 0
3. No optimal solution x1, x2, x3, x4 Ú 0

5.1 The Table Method: An Introduction to the Simplex Method EXAMPLES
• A linear programming problem is said to be a standard maximization problem in standard form if its
mathematical model is of the following form: Maximize the objective function
P = c1x1 + c2x2 + g + ckxk

subject to problem constraints of the form
a1x1 + a2x2 + g + akxk … b b Ú 0
with nonnegative constraints
x1, x2, c , xk Ú 0
• The system of inequalities (i-system) of a linear programming problem is converted to a system of equa- Ex. 1, p. 269
tions (e-system) by means of slack variables. A solution of the e-system is a feasible solution if the val- Ex. 2, p. 270
ues of all decision variables and slack variables are nonnegative. The feasible solutions of the e-system
Ex. 3, p. 271
correspond to the points in the feasible region of the i-system. A basic solution of the e-system is found
by setting k of the variables equal to 0, where k is the number of decision variables x1, x2, c , xk.
A solution of the e-system that is both basic and feasible is called a basic feasible solution. The table
method for solving a linear programming problem consists of constructing a table of all basic solutions,
determining which of the basic solutions are feasible, and then maximizing the objective function over the
basic feasible solutions. A procedure for carrying out the table method in the case of k = 2 decision vari-
ables is given on page 268. For an arbitrary number of decision variables, see the procedure on page 274.
• The fundamental theorem of linear programming can be formulated in terms of basic feasible solu-
tions. It states that an optimal solution to the linear programming problem, if one exists, must occur at
one or more of the basic feasible solutions.
• The k variables that are assigned the value 0, in order to generate a basic solution, are called nonbasic Ex. 4, p. 273
variables. The remaining variables are called basic variables. So the classification of variables as basic
or nonbasic depends on the basic solution under consideration.
• The benefit of the table method is that it gives a procedure for finding all corner points of the feasible
region without drawing a graph. But the table method has a drawback: If the number of decision vari-
ables and problem constraints is large, then the number of rows in the table (that is, the number of basic
solutions) becomes too large for practical implementation. The simplex method, discussed in Section 5.2,
gives a practical method for solving large linear programming problems.
5.2 The Simplex Method: Maximization with Problem Constraints of the Form …
• Adding the objective function to the system of constraint equations produces the initial system. Nega-
tive values of the objective function variable are permitted in a basic feasible solution as long as all
other variables are nonnegative. The fundamental theorem of linear programming also applies to initial
systems.
• The augmented matrix of the initial system is called the initial simplex tableau. The simplex method Ex. 1, p. 285
consists of performing pivot operations, starting with the initial simplex tableau, until an optimal solu- Ex. 2, p. 287
tion is found (if one exists). The procedure is illustrated in Figure 2 (p. 285). Ex. 3, p. 288
5.3 The Dual Problem: Minimization with Problem Constraints of the Form Ú
• By the Fundamental Principle of Duality, a linear programming problem that asks for the minimum of Ex. 1, p. 296
the objective function over a region described by Ú problem constraints can be solved by first forming Ex. 2, p. 299
the dual problem and then using the simplex method. Ex. 3, p. 301
5.4 Maximization and Minimization with Mixed Problem Constraints
• The big M method can be used to find the maximum of any objective function on any feasible region. Ex. 1, p. 312
The solution process involves the introduction of two new types of variables, surplus variables and
artificial variables, and a modification of the objective function. The result is an initial tableau that can
be transformed into the tableau of a modified problem.
• Applying the simplex method to the modified problem produces a solution to the original problem, if one Ex. 2, p. 314
exists. Ex. 3, p. 315
• The dual method can be used to solve only certain minimization problems. But all minimization prob- Ex. 4, p. 317
lems can be solved by using the big M method to find the maximum of the negative of the objective Ex. 5, p. 319
function. The big M method also lends itself to computer implementation.
Review Exercises
Work through all the problems in this chapter review and check 6. Describe geometrically the set of points in the plane such that
your answers in the back of the book. Answers to all review prob- s1 6 0.
7. Use the basic feasible solutions to maximize
P = 50x1 + 60x2.
8. A linear programming problem has 6 decision variables and
A Problems 1–7 refer to the partially completed table of the six
3 problem constraints. How many rows are there in the table
basic solutions to the e-system
of basic solutions of the corresponding e-system?
2x1 + 5x2 + s1 = 32 9. Given the linear programming problem
x1 + 2x2 + s2 = 14 Maximize P = 6x1 + 2x2
x1 x2 s1 s2 subject to 2x1 + x2 … 8
(A) 0 0 32 14 x1 + 2x2 … 10
(B) 0 6.4 0 1.2 x1, x2 Ú 0
(C) 0 7 -3 0
convert the problem constraints into a system of equations
(D) 16 0 0 -2 using slack variables.
(E) 0 0 10. How many basic variables and how many nonbasic variables
(F) 0 0 are associated with the system in Problem 9?
1. In basic solution (B), which variables are basic? 11. Find all basic solutions for the system in Problem 9, and
determine which basic solutions are feasible.
2. In basic solution (D), which variables are nonbasic?
12. Write the simplex tableau for Problem 9, and circle the pivot
3. Find basic solution (E).
element. Indicate the entering and exiting variables.
4. Find basic solution (F).
13. Solve Problem 9 using the simplex method.
5. Which of the six basic solutions are feasible?
14. For the simplex tableau below, identify the basic and non- 23. Solve Problem 22 by applying the simplex method to the dual
basic variables. Find the pivot element, the entering and exit- problem.
ing variables, and perform one pivot operation.
Solve the linear programming Problems 24 and 25.
x1 x2 x3 s1 s2 s3 P
24. Maximize P = 5x1 + 3x2 - 3x3
2 1 3 -1 0 0 0 20
subject to x1 - x2 - 2x3 … 3
3 0 4 1 1 0 0 30
D 4 T 2x1 + 2x2 - 5x3 … 10
2 0 5 2 0 1 0 10
-8 0 -5 3 0 0 1 50 x1, x2, x3 Ú 0
15. Find the basic solution for each tableau. Determine whether 25. Maximize P = 5x1 + 3x2 - 3x3
the optimal solution has been reached, additional pivoting is subject to x1 - x2 - 2x3 … 3
required, or the problem has no optimal solution.
x1 + x2 … 5
x1 x2 s1 s2 P x1, x2, x3 Ú 0
(A)
4 1 0 0 0 2
C 2 0 1 1 0 3 5S 26. Solve the linear programming problem using the table
method:
-2 0 3 0 1 12
Maximize P = 10x1 + 7x2 + 8x3
x1 x2 s1 s2 P
(B) subject to 2x1 + x2 + 3x3 … 12
-1 3 0 1 0 7
C 0 2 1 0 0 3 0S x1, x2 Ú 0
-2 1 0 0 1 22 27. Refer to Problem 26. How many pivot operations are required
x1 x2 s1 s2 P to solve the linear programming problem using the simplex
(C) method?
1 -2 0 4 0 6
C0 2 1 6 0 3 15 S In Problems 28 and 29,
0 3 0 2 1 10
(A) Introduce slack, surplus, and artificial variables and form the
16. Form the dual problem of modified problem.
Minimize C = 5x1 + 2x2 (B) Write the preliminary simplex tableau for the modified prob-
lem and find the initial simplex tableau.
subject to x1 + 3x2 Ú 15
(C) Find the optimal solution of the modified problem by apply-
2x1 + x2 Ú 20 ing the simplex method to the initial simplex tableau.
x1, x2 Ú 0 (D) Find the optimal solution of the original problem, if it exists.
17. Write the initial system for the dual problem in Problem 16.
28. Maximize P = x1 + 3x2
18. Write the first simplex tableau for the dual problem in Prob- subject to x1 + x2 Ú 6
lem 16 and label the columns.
x1 + 2x2 … 8
19. Use the simplex method to find the optimal solution of the
x1, x2 Ú 0
dual problem in Problem 16.
20. Use the final simplex tableau from Problem 19 to find the opti- 29. Maximize P = x1 + x2
mal solution of the linear programming problem in Problem 16. subject to x1 + x2 Ú 5
B 21. Solve the linear programming problem using the simplex method. x1 + 2x2 … 4
Maximize P = 3x1 + 4x2 x1, x2 Ú 0
subject to 2x1 + 4x2 … 24 30. Find the modified problem for the following linear program-
3x1 + 3x2 … 21 ming problem. (Do not solve.)
4x1 + 2x2 … 20 Maximize P = 2x1 + 3x2 + x3
x1, x2 Ú 0
subject to x1 - 3x2 + x3 … 7
22. Form the dual problem of the linear programming problem - x1 - x2 + 2x3 … - 2
3x1 + 2x2 - x3 = 4
subject to x1 + x2 Ú 10
x1, x2, x3 Ú 0
x1 + 2x2 Ú 15
x2 Ú 3
x1, x2 Ú 0
Write a brief verbal description of the type of linear programming (A) If the oil stock yields 12% and the steel stock yields 9%,
problem that can be solved by the method indicated in Problems how much money should be invested in each alternative
31–33. Include the type of optimization, the number of variables, in order to maximize the return? What is the maximum
the type of constraints, and any restrictions on the coefficients and return?
constants.
(B) Repeat part (A) if the oil stock yields 9% and the steel
31. Basic simplex method with slack variables stock yields 12%.
32. Dual problem method 39. Manufacturing. A company manufactures outdoor furni-
33. Big M method ture consisting of regular chairs, rocking chairs, and chaise
lounges. Each piece of furniture passes through three dif-
C 34. Solve the following linear programming problem by the ferent production departments: fabrication, assembly, and
simplex method, keeping track of the obvious basic solution at finishing. Each regular chair takes 1 hour to fabricate, 2 hours
each step. Then graph the feasible region and illustrate the path to assemble, and 3 hours to finish. Each rocking chair takes
to the optimal solution determined by the simplex method. 2 hours to fabricate, 2 hours to assemble, and 3 hours to fin-
Maximize P = 2x1 + 3x2 ish. Each chaise lounge takes 3 hours to fabricate, 4 hours to
assemble, and 2 hours to finish. There are 2,500 labor-hours
available in the fabrication department, 3,000 in the assembly
3x1 + x2 … 26 department, and 3,500 in the finishing department. The com-
x1 … 8 pany makes a profit of $17 on each regular chair, $24 on each
x2 … 10 rocking chair, and $31 on each chaise lounge.
x1, x2 Ú 0 (A) How many chairs of each type should the company pro-
duce in order to maximize profit? What is the maximum
35. Solve by the dual problem method: profit?
Minimize C = 3x1 + 2x2 (B) Discuss the effect on the optimal solution in part (A) if
subject to 2x1 + x2 … 20 the profit on a regular chair is increased to $25 and all
2x1 + x2 Ú 9 other data remain the same.
x1 + x2 Ú 6 (C) Discuss the effect on the optimal solution in part (A)
x1, x2 Ú 0 if the available hours on the finishing department are
reduced to 3,000 and all other data remain the same.
36. Solve Problem 35 by the big M method.
40. Shipping schedules. A company produces motors for
37. Solve by the dual problem method: washing machines at factory A and factory B. The motors are
then shipped to either plant X or plant Y, where the washing
Minimize C = 15x1 + 12x2 + 15x3 + 18x4
machines are assembled. The maximum number of motors
subject to x1 + x2 … 240 that can be produced at each factory monthly, the minimum
x3 + x4 … 500 number required monthly for each plant to meet anticipated
x1 + x3 Ú 400 demand, and the shipping charges for one motor are given in
the table. Determine a shipping schedule that will minimize
x2 + x4 Ú 300
the cost of transporting the motors from the factories to the
x1, x2, x3, x4 Ú 0 assembly plants.
Maximum
Plant X Plant Y Production
Applications Factory A $5 $8 1,500

Factory B $9 $7 1,000
In Problems 38–41, construct a mathematical model in the form Minimum Requirement 900 1,200
of a linear programming problem. (The answers in the back
of the book for these application problems include the model.) 41. Blending–food processing. A company blends long-grain
Then solve the problem by the simplex, dual problem, or big rice and wild rice to produce two brands of rice mixes: brand
M methods. A, which is marketed under the company’s name, and brand
38. Investment. An investor has $150,000 to invest in oil stock, B, which is marketed as a generic brand. Brand A must
steel stock, and government bonds. The bonds are guaranteed contain at least 10% wild rice, and brand B must contain at
to yield 5%, but the yield for each stock can vary. To protect least 5% wild rice. Long-grain rice costs $0.70 per pound,
against major losses, the investor decides that the amount and wild rice costs $3.40 per pound. The company sells brand
invested in oil stock should not exceed $50,000 and that the A for $1.50 a pound and brand B for $1.20 a pound. The com-
total amount invested in stock cannot exceed the amount pany has 8,000 pounds of long-grain rice and 500 pounds of
invested in bonds by more than $25,000. wild rice on hand. How should the company use the available
rice to maximize its profit? What is the maximum profit?
6 Logic, Sets, and
Counting
6.1 Logic
Introduction
6.2 Sets
Quality control is crucial in manufacturing to insure product safety and reliabil-
6.3 Basic Counting ity. One method of quality control is to test a sample of manufactured parts (see
Principles Problems 71 and 72 in Section 6.4). If the sample does not pass the test, there
6.4 Permutations and is a problem in production that must be rectified. The mathematics behind this
Combinations method of quality control involves logic, sets, and counting, the key concepts
explored in Chapter 6.
Logic and sets form the foundation of mathematics. That foundation remained
largely out of view in Appendix A and Chapters 1 through 5. In Chapter 6, how-
ever, we study logic and sets explicitly with an eye toward the topic of probability
in Chapter 7. We introduce the symbolic logic of propositions in Section 6.1,
study sets in Section 6.2, and consider various counting techniques in Section 6.4.
329
330 CHAPTER 6 Logic, Sets, and Counting
6.1 Logic
■■ Propositions and Connectives Consider the kind of logical reasoning that is used in everyday language. For
■■ Truth Tables
example, suppose that the following two statements are true:
■■ Logical Implications and “If today is Saturday, then Derek plays soccer” and
Equivalences “Today is Saturday.”
From the truth of these two statements, we can conclude that
“Derek plays soccer.”
Similarly, suppose that the following two mathematical statements are true:
“If the sum of the digits of 71,325 is divisible by 9, then 71,325 is divis-
ible by 9” and
“The sum of the digits of 71,325 is divisible by 9.”
From the truth of these two statements, we can conclude that
“71,325 is divisible by 9.”
Logic is the study of the form of arguments. Each of the preceding arguments,
the first about Derek and the second about divisibility by 9, has the same form,
namely,
31p S q2 ¿ p4 1 q
In this section, we introduce the notation that is required to represent an argument in
compact form, and we establish precision in the use of logical deduction that forms
the foundation for the proof of mathematical theorems.
Propositions and Connectives

A proposition is a statement (not a question or command) that is either true or false.
So the statement
“There is a prime number between 2,312 and 2,325”
is a proposition. It is a proposition even if we do not know or cannot determine
whether it is true. We use lowercase letters such as p, q, and r to denote propositions.
If p and q are propositions, then the compound propositions
¬ p, p ¡ q, p ¿ q, and p S q
can be formed using the negation symbol ¬ and the connectives ¡ , ¿ , and S . These
propositions are called “not p,” “p or q,” “p and q,” and “if p then q,” respectively.
We use a truth table to specify each of these compound propositions. A truth table
gives the proposition’s truth value, T (true) or F (false), for all possible values of its
variables.
DEFINITION Negation
If p is a proposition, then the proposition ¬ p, read not p, or the negation of p, is
false if p is true and true if p is false.
p ¬p
T F
F T
SECTION 6.1 Logic 331
DEFINITION Disjunction
If p and q are propositions, then the proposition p ¡ q, read p or q, or the disjunction
of p and q, is true if p is true, or if q is true, or if both are true, and is false otherwise.
p q p ¡ q
T T T
T F T
F T T
F F F
Note that or is used in the inclusive sense; that is, it includes the possibility that
both p and q are true. This mathematical usage differs from the way that or is some-
times used in everyday language (“I will order chicken or I will order fish”) when we
intend to exclude the possibility that both are true.
DEFINITION Conjunction
If p and q are propositions, then the proposition p ¿ q, read p and q, or the
conjunction of p and q, is true if both p and q are true and is false otherwise.
p q p ¿ q
T T T
T F F
F T F
F F F
The truth table for p ¿ q is just what you would expect, based on the use of
“and” in everyday language. Note that there is just one T in the third column of the
truth table for conjunction; both p and q must be true in order for p ¿ q to be true.
DEFINITION Conditional
If p and q are propositions, then the proposition p S q, read if p then q, or the
conditional with hypothesis p and conclusion q, is false if p is true and q is false,
but is true otherwise.
p q pSq
T T T
T F F
F T T
F F T
Note that the definition of p S q differs somewhat from the use of “if p then q”
in everyday language. For example, we might question whether the proposition
“If Paris is in Switzerland, then Greenwich is in London”
is true on the grounds that there is no apparent connection between p (“Paris is in
Switzerland”) and q (“Greenwich is in London”). We consider it to be true, however,
in accordance with the definition of p S q, because p is false. Whenever the hypoth-
esis p is false, we say that the conditional p S q is vacuously true.
CONCEPTUAL INSIGHT
It is helpful to think of the conditional as a guarantee. For example, an instruc-
tor of a mathematics course might give a student the guarantee: “If you score at
least 90%, then you will get an A.” Suppose the student scores less than 90%,
so the hypothesis is false. The guarantee remains in effect even though it is not
applicable. We say that the conditional statement is vacuously true. In fact, there
is only one circumstance in which the conditional statement could be false: The
student scores at least 90% (that is, the hypothesis is true), but the grade is not an
A (the conclusion is false).
EXAMPLE 1 Compound Propositions Consider the propositions p and q:

p: “4 + 3 is even.”
q: “42 + 32 is odd.”
Express each of the following propositions as an English sentence and determine
whether it is true or false.
(A) ¬ p (B) ¬ q (C) p ¡ q (D) p ¿ q (E) p S q
SOLUTION
(A) ¬ p: “4 + 3 is not even”
(Note that we modified the wording of “Not 4 + 3 is even” to standard English
usage.) Because 4 + 3 = 7 is odd, p is false, and therefore ¬p is true.
(B) ¬q: “42 + 32 is not odd”
Because 42 + 32 = 25 is odd, q is true, and therefore ¬q is false.
(C) p ¡ q: “4 + 3 is even or 42 + 32 is odd”
Because q is true, p ¡ q is true.
(D) p ¿ q: “4 + 3 is even and 42 + 32 is odd”
Because p is false, p ¿ q is false.
(E) p S q: “if 4 + 3 is even, then 42 + 32 is odd”
Because p is false, p S q is (vacuously) true.
Matched Problem 1 Consider the propositions p and q:

p: “142 6 200”
q: “232 6 500”
(A) ¬ p (B) ¬ q (C) p ¡ q (D) p ¿ q (E) p S q
With any conditional p S q, we associate two other propositions: the converse

of p S q and the contrapositive of p S q.
DEFINITION Converse and Contrapositive

Let p S q be a conditional proposition. The proposition q S p is called the converse
of p S q. The proposition ¬ q S ¬ p is called the contrapositive of p S q.
EXAMPLE 2 Converse and Contrapositive Consider the propositions p and q:

p: “2 + 2 = 4”
q: “9 is a prime”

(A) p S q (B) The converse of p S q (C) The contrapositive of p S q
SOLUTION
(A) p S q: “if 2 + 2 = 4, then 9 is a prime”
Because p is true and q is false, p S q is false.
(B) q S p: “if 9 is a prime, then 2 + 2 = 4”
Because q is false, q S p is vacuously true.
(C) ¬ q S ¬ p: “if 9 is not prime, then 2 + 2 is not equal to 4”
Because q is false and p is true, ¬ q is true and ¬ p is false, so ¬ q S ¬ p is false.
Matched Problem 2 Consider the propositions p and q:

p: “52 + 122 = 132”
q: “72 + 242 = 252”
(A) p S q (B) The converse of p S q (C) The contrapositive of p S q
Truth Tables
A truth table for a compound proposition specifies whether it is true or false for any
assignment of truth values to its variables. Such a truth table can be constructed for
any compound proposition by referring to the truth tables in the definitions of ¬, ¡ ,
¿ , and S .
EXAMPLE 3 Constructing Truth Tables Construct the truth table for ¬p ¡ q.

SOLUTION The proposition contains two variables, p and q, so the truth table will
consist of four rows, one for each possible assignment of truth values to two variables
p q ¬p ¬p ¡ q (TT, TF, FT, FF). Although the truth table itself consists of three columns, one labeled
T T F T
p, another labeled q, and the third labeled ¬ p ¡ q, it is helpful to insert an additional
T F F F
column labeled ¬ p. The entries in that additional column are obtained from the first
F T T T column, changing any T to F, and vice versa, in accordance with the definition of ¬.
F F T T The entries in the last column are obtained from the third and second columns, in ac-
cordance with the definition of ¡ (in a given row, if either entry in those columns is a
T, then the entry in the last column is T; if both are F, the entry in the last column is F).
Note that the truth table for ¬ p ¡ q is identical to the truth table for p S q (see
the definition of a conditional).
Matched Problem 3 Construct the truth table for p ¿ ¬ q.
EXAMPLE 4 Constructing a Truth Table Construct the truth table for 31p S q2 ¿ p4 S q.
SOLUTION It is helpful to insert a third column labeled p S q and a fourth labeled
1p S q2 ¿ p. We complete the first three columns. Then we use the third and first
columns to complete the fourth, and we use the fourth and second columns to com-
plete the last column.
p q pSq 1p S q2 ¿ p 31p S q2 ¿ p4 S q
T T T T T
T F F F T
F T T F T
F F T F T
Note that 31p S q2 ¿ p4 S q is always true, regardless of the truth values of

p and q.
Matched Problem 4 Construct the truth table for 31p S q2 ¿ ¬ q4 S ¬ p.
Any proposition is either a tautology, a contradiction, or a contingency. The

proposition 31p S q2 ¿ p4 S q of Example 4, which is always true, is a tautology.
The proposition ¬ p ¡ q of Example 3, which may be true or false, is a contingency.
DEFINITION Tautology, Contradiction, and Contingency

A proposition is a tautology if each entry in its column of the truth table is T, a
contradiction if each entry is F, and a contingency if at least one entry is T and at
least one entry is F.
EXAMPLE 5 Constructing a Truth Table Construct the truth table for p ¿ ¬1p ¡ q2.
SOLUTION It is helpful to insert a third column labeled p ¡ q and a fourth labeled
¬1p ¡ q2. We complete the first three columns. Then we use the third column to
complete the fourth, and we use the first and fourth columns to complete the last
column.
p q p ¡ q ¬1p ¡ q2 p ¿ ¬1p ¡ q2
T T T F F
T F T F F
F T T F F
F F F T F
Note that p ¿ ¬1p ¡ q2 is a contradiction; it is always false, regardless of the truth

values of p and q.
Matched Problem 5 Construct the truth table for 1p S q2 ¿ 1p ¿ ¬q2.
The LOGIC menu on the TI-84 Plus CE contains the operators “and,” “or,” “xor,” and
“not.” The truth table for p ¿ q, for example, can be displayed by entering all com-
binations of truth values for p and q in lists L 1 and L 2 (using 1 to represent T and 0 to
represent F), and then, for L 3, entering “L 1 and L 2” (see Fig. 1).
Figure 1
(A) Use a calculator to display the truth table for the “exclusive or” operator “xor.”
(B) Show that the truth table for “xor” is identical to the truth table for the proposition
1p ¡ q2 ¿ ¬1p ¿ q2.
Logical Implications and Equivalences

At the beginning of the section, we noted that the notation 31p S q2 ¿ p4 1 q
could be used to represent the form of a familiar logical deduction. The notion of
such a deduction can be defined in terms of truth tables. Consider the truth tables for
the propositions 1p S q2 ¿ p and q:
Table 1
p q pSq 1p S q2 ¿ p
T T T T
T F F F
F T T F
F F T F
Whenever the proposition 1p S q2 ¿ p is true (in this case, in the first row
only), the proposition q is also true. We say that 1p S q2 ¿ p implies q, or that
31p S q2 ¿ p4 1 q is a logical implication.
DEFINITION Logical Implication

Consider the rows of the truth tables for the compound propositions P and Q. If
whenever P is true, Q is also true, we say that P logically implies Q and write
P 1 Q. We call P 1 Q a logical implication.
CONCEPTUAL INSIGHT
If P 1 Q, then whenever P is true, so is Q. So to say that P implies Q is the same
as saying that the proposition P S Q is a tautology. This gives us two methods of
verifying that P implies Q: We can check the rows of the truth tables for P and Q
as we did previously, or we can construct the truth table for P S Q to check
that it is a tautology. Compare the truth tables in Table 1 with the truth table of
Example 4 to decide which of the two methods you prefer.
EXAMPLE 6 Verifying a Logical Implication Show that 31p S q2 ¿ ¬ q4 1 ¬ p.

SOLUTION To construct the truth table for 1p S q2 ¿ ¬q, it is helpful to insert
a third column labeled p S q and a fourth labeled ¬q. We complete the first three
columns. Then we use the second column to complete the fourth, and we use the
third and fourth columns to complete the fifth column.
p q pSq ¬q 1p S q2 ¿ ¬q ¬p
T T T F F F
T F F T F F
F T T F F T
F F T T T T
Now we compare the fifth column to the sixth: Whenever 1p S q2 ¿ ¬ q is true

(in the fourth row only), ¬ p is also true. We conclude that 31p S q2 ¿ ¬ q4 1 ¬ p.
Matched Problem 6 Show that 31p S q2 S p24 1 1q S p2.
If the compound propositions P and Q have identical truth tables, then P 1 Q

and Q 1 P. In this case, we say that P and Q are logically equivalent.
DEFINITION Logical Equivalence

If the compound propositions P and Q have identical truth tables, we say that P and
Q are logically equivalent and write P K Q. We call P K Q a logical equivalence.
In Example 3, we noted that p S q and ¬ p ¡ q have identical truth tables.

Therefore, p S q K ¬ p ¡ q. This is formula (4) in Table 2, which lists several
logical equivalences. The first three equivalences of Table 2 are obvious, and the last
three equivalences are established in Example 7, Matched Problem 7, and Example 8,
respectively.
Table 2 Some Logical Equivalences

¬1¬ p2 K p (1)
p ¡ q K q ¡ p (2)
p ¿ q K q ¿ p (3)
p S q K ¬p ¡ q (4)
¬1p ¡ q2 K ¬ p ¿ ¬ q (5)
¬1p ¿ q2 K ¬ p ¡ ¬ q (6)
p S q K ¬q S ¬p (7)
CONCEPTUAL INSIGHT
Formulas (5) and (6) in Table 2 are known as De Morgan’s laws. They may
remind you of the way a negative sign is distributed over a binomial in algebra:
- 1a + b2 = 1 - a2 + 1 - b2
But there is an important difference: Formula (5) has a disjunction 1 ¡ 2 on the left
side but a conjunction 1 ¿ 2 on the right side. Similarly, formula (6) has a conjunc-
tion on the left side but a disjunction on the right side.
EXAMPLE 7 Verifying a Logical Equivalence Show that ¬ 1p ¡ q2 K ¬ p ¿ ¬ q.

SOLUTION We construct truth tables for ¬1p ¡ q2 and ¬ p ¿ ¬ q.
p q p ¡ q ¬1 p ¡ q2 ¬p ¬q ¬p ¿ ¬q
T T T F F F F
T F T F F T F
F T T F T F F
F F F T T T T
The fourth and seventh columns are identical, so ¬1p ¡ q2 K ¬p ¿ ¬q.
Matched Problem 7 Show that ¬1p ¿ q2 K ¬ p ¡ ¬ q.
One way to show that two propositions are logically equivalent is to check that
their truth tables are identical (as in Example 7). Another way is to convert one to the
other by a sequence of steps, where a known logical equivalence is used at each step
to replace part or all of the proposition by an equivalent proposition. This procedure,
analogous to simplifying an algebraic expression, is illustrated in Example 8 to show
that any conditional proposition is logically equivalent to its contrapositive.
EXAMPLE 8 Any Conditional and Its Contrapositive Are Logically Equivalent Show that
p S q K ¬ q S ¬ p.
SOLUTION Each step is justified by a reference to one of the formulas in Table 2:
pSq K ¬p ¡ q By (4)
K q ¡ ¬p By (2)
K ¬1¬q2 ¡ ¬ p By (1)
K ¬q S ¬p By (4)
Therefore, p S q K ¬ q S ¬ p.
Matched Problem 8 Use equivalences from Table 2 to show that

p S q K ¬1¬ q ¿ p2.
If a compound proposition contains three variables p, q, and r, then its truth table will
have eight rows, one for each of the eight ways of assigning truth values to p, q, and
r (TTT, TTF, TFT, TFF, FTT, FTF, FFT, FFF). Construct truth tables to verify the fol-
lowing logical implication and equivalences:
(A) 1p S q2 ¿ 1q S r2 1 1p S r2
(B) p ¿ 1q ¡ r2 K 1p ¿ q2 ¡ 1p ¿ r2
(C) p ¡ 1q ¿ r2 K 1p ¡ q2 ¿ 1p ¡ r2
Exercises 6.1
W Skills Warm-up Exercises 13. The converse of p S q
In Problems 1–6, refer to the footnote for the definitions of divisor, 14. The contrapositive of p S q
multiple, prime, even, and odd.*
In Problems 15–20, express each proposition as an English sen-
1. List the positive integers that are divisors of 20. tence and determine whether it is true or false, where r and s are
2. List the positive integers that are divisors of 24. the propositions
r: “the moon is a cube”
3. List the positive multiples of 11 that are less than 60. s: “rain is wet”
4. List the positive multiples of 17 that are less than 100. 15. r S s 16. r ¿ s
5. List the primes between 20 and 30. 17. r ¡ s 18. ¬ r
6. List the primes between 10 and 20. 19. The contrapositive of r S s
7. Explain why the sum of any two odd integers is even. 20. The converse of r S s
8. Explain why the product of any two odd integers is odd.
In Problems 21–28, describe each proposition as a negation, dis-
junction, conjunction, or conditional, and determine whether the
A In Problems 9–14, express each proposition as an English sen- proposition is true or false.
tence and determine whether it is true or false, where p and q are
the propositions 21. - 3 6 0 or - 3 7 0
p: “91 is prime” q: “91 is odd” 22. - 3 6 0 and - 3 7 0
9. ¬ p 10. p ¡ q 23. If - 3 6 0, then 1 - 32 2 6 0
11. p ¿ q 12. p S q 24. - 3 is not greater than 0
*An integer d is a divisor of an integer n (and n is a multiple of d) if n = kd 25. 11 is not prime

for some integer k. An integer n is even if 2 is a divisor of n; otherwise, n 26. 35 is prime and 35 is odd
is odd. An integer p 7 1 is prime if its only positive divisors are 1 and p.
27. 35 is prime or 35 is odd 67. ¬1p S q2 K p ¿ ¬ q

28. If 4 is even, then 4 is prime 68. ¬1¬ p S ¬ q2 K q ¿ ¬ p
In Problems 29–34, state the converse and the contrapositive of 69. If p is the proposition “I am hungry” and q is the proposition
the given proposition. “I am thirsty,” rewrite the sentence “I am hungry, but I am not
thirsty” using symbols.
29. If triangle ABC is equilateral, then triangle ABC is equiangu-
lar. 70. Let p be the proposition “every politician is honest.” Explain
why the statement “every politician is dishonest” is not
30. If triangle ABC is isosceles, then the base angles of triangle equivalent to ¬ p. Express ¬ p as an English sentence without
ABC are congruent. using the word not.
31. If f1x2 is a linear function with negative slope, then f1x2 is 71. If the conditional proposition p is a contingency, is ¬ p a
a decreasing function. contingency, a tautology, or a contradiction? Explain.
32. If g1x2 is a quadratic function, then g1x2 is a function that is 72. If the conditional proposition p is a contradiction, is ¬ p a
neither increasing nor decreasing. contingency, a tautology, or a contradiction? Explain.
33. If n is an integer that is a multiple of 8, then n is an integer 73. Can a conditional proposition be false if its contrapositive is
that is a multiple of 2 and a multiple of 4. true? Explain.
34. If n is an integer that is a multiple of 10, then n is an integer 74. Can a conditional proposition be false if its converse is true?
that is a multiple of 2 and a multiple of 5. Explain.
B In Problems 35–52, construct a truth table for the proposition and
determine whether the proposition is a contingency, tautology, or
contradiction. 1. (A) “142 is not less than 200”; false
(B) “232 is not less than 500”; true
35. ¬ p ¿ q 36. p ¡ ¬ q
(C) “142 is less than 200 or 232 is less than 500”; true
37. ¬ p S q 38. p S ¬ q (D) “142 is less than 200 and 232 is less than 500”; false
(E) “If 142 is less than 200, then 232 is less than 500”; false
39. q ¿ 1p ¡ q2 40. q ¡ 1p ¿ q2
2. (A) “If 52 + 122 = 132, then 72 + 242 = 252”; true
41. p ¡ 1p S q2 42. p ¿ 1p S q2 (B) “If 72 + 242 = 252, then 52 + 122 = 132; true
(C) “If 72 + 242 ∙ 252, then 52 + 122 ∙ 132; true
43. p S 1p ¿ q2 44. p S 1p ¡ q2
3. p q p ¿ ¬q
45. 1p S q2 S ¬ p 46. 1p S q2 S ¬ q
T T F
47. ¬ p S 1p ¡ q2 48. ¬ p S 1p ¿ q2
T F T
49. 1 p S q2 ¡ 1q S p2 50. 1p S q2 ¿ p F T F
51. 1¬ p ¿ q2 ¿ 1q S p2 52. 1p S ¬ q2 ¿ 1p ¿ q2 F F F
4. p q 31p S q2 ¿ ¬q4 S ¬p
C In Problems 53–58, construct a truth table to verify each implication.
T T T
53. p 1 p ¡ q 54. ¬ p 1 p S q
T F T
55. 1 p ¿ q2 1 1 p ¡ q2 56. p ¿ q 1 p S q F T T
57. ¬ p S 1q ¿ ¬ q2 1 p 58. 1p ¿ ¬ p2 1 q F F T
In Problems 59–64, construct a truth table to verify each 5. p q 1p S q2 ¿ 1p ¿ ¬q2

equivalence. T T F
59. ¬p S 1p ¡ q2 K p ¡ q T F F
60. q S 1¬ p ¿ q2 K ¬1p ¿ q2 F T F
F F F
61. q ¿ 1p ¡ q2 K q ¡ 1p ¿ q2
6. p q 1p S q2 S p qSp
62. p ¿ 1p S q2 K p ¿ q
T T T T
63. p ¡ 1p S q2 K p S 1p ¡ q2
T F T T
64. p S 1p ¿ q2 K p S q F T F F
In Problems 65–68, verify each equivalence using formulas from F F F T
Table 2.
Whenever 1p S q2 S p is true, so is q S p, so
65. p S ¬ q K ¬ 1p ¿ q2 31p S q2 S p2 4 1 1q S p2.
66. ¬p S q K p ¡ q
SECTION 6.2 Sets 339
7. p q ¬1p ¿ q2 ¬p ¡ ¬q 8. p S q K ¬ p ¡ q By (4)
K q ¡ ¬p By (2)
T T F F
K ¬1¬q2 ¡ ¬ p By (1)
T F T T
K ¬1¬ q ¿ p2 By (6)
F T T T
F F T T
The third and fourth columns are identical, so

¬1p ¿ q2 K ¬p ¡ ¬q.
6.2 Sets
■■ Set Properties and Set Notation In this section, we review a few key ideas about sets. Set concepts and notation help
■■ Venn Diagrams and Set Operations
us talk about certain mathematical ideas with greater clarity and precision, and they
are indispensable to a clear understanding of probability.
■■ Application
Set Properties and Set Notation
We can think of a set as any collection of objects specified in such a way that we can
tell whether any given object is or is not in the collection. Capital letters, such as A,
B, and C, are often used to designate particular sets. Each object in a set is called a
member, or element, of the set. Symbolically,
a ∊ A means “a is an element of set A”
a F A means “a is not an element of set A”
A set without any elements is called the empty, or null, set. For example, the set
of all people over 20 feet tall is an empty set. Symbolically,
∅ denotes the empty set
A set is described either by listing all its elements between braces 56 (the listing
method) or by enclosing a rule within braces that determines the elements of the set
(the rule method). So if P1x2 is a statement about x, then
S = 5x ∙ P1x26 means “S is the set of all x such that P1x2is true“
Recall that the vertical bar within the braces is read “such that.” The following
example illustrates the rule and listing methods of representing sets.
EXAMPLE 1 Representing Sets

Rule method Listing method
5x ∙ x is a weekend day6 = 5Saturday, Sunday6
5x ∙ x2 = 46 = 5 - 2, 26
5x ∙ x is an odd positive counting number6 = 51, 3, 5, c6
The three dots 1c2 in the last set of Example 1 indicate that the pattern
established by the first three entries continues indefinitely. The first two sets in
Example 1 are finite sets (the elements can be counted, and there is an end); the
last set is an infinite set (there is no end when counting the elements). When listing
the elements in a set, we do not list an element more than once, and the order in
which the elements are listed does not matter.
Matched Problem 1 Let G be the set of all numbers such that x2 = 9.

(A) Denote G by the rule method.
(B) Denote G by the listing method.
(C) Indicate whether the following are true or false: 3 ∊ G, 9 F G.
If each element of a set A is also an element of set B, we say that A is a subset of

B. For example, the set of all women students in a class is a subset of the whole class.
Note that the definition implies that every set is a subset of itself. If set A and set B
have exactly the same elements, then the two sets are said to be equal. Symbolically,
A⊂B means “A is a subset of B”
A = B means “A and B have exactly the same elements”
A⊄B means “A is not a subset of B”
A ∙ B means “A and B do not have exactly the same elements”
From the definition of subset, we conclude that
∅ is a subset of every set, and
if A ⊂ B and B ⊂ A, then A = B.
CONCEPTUAL INSIGHT
To conclude that ∅ is a subset of any set A, we must show that the conditional
proposition “if x ∊ ∅, then x ∊ A” is true. But the set ∅ has no element, so the
hypothesis of the conditional is false, and therefore the conditional itself is vacu-
ously true (see Section 6.1). We correctly conclude that ∅ is a subset of every set.
Note, however, that ∅ is not an element of every set.
EXAMPLE 2 Set Notation If A = 5 - 3, - 1, 1, 36, B = 53, - 3, 1, - 16, and C = 5 - 3, - 2,

- 1, 0, 1, 2, 36, then each of the following statements is true:
A = B A⊂C A⊂B
C ∙ A C⊄A B⊂A
∅⊂A ∅⊂C ∅FA
Matched Problem 2 Given A = 50, 2, 4, 66, B = 50, 1, 2, 3, 4, 5, 66,
and C = 52, 6, 0, 46, indicate whether the following relationships are true (T) or
false (F):
(A) A ⊂ B (B) A ⊂ C (C) A = C
(D) C ⊂ B (E) B ⊄ A (F) ∅ ⊂ B
EXAMPLE 3 Subsets List all subsets of the set 5a, b, c6.

SOLUTION
5a, b, c6, 5a, b6, 5a, c6, 5b, c6, 5a6, 5b6, 5c6, ∅
Matched Problem 3 List all subsets of the set 51, 26.
Which of the following statements are true?

(A) ∅ ⊂ ∅
(B) ∅ ∊ ∅
(C) ∅ = 506
(D) ∅ ⊂ 506
U
Venn Diagrams and Set Operations
A If the set U is the set of all rental units in a city, and A is the set of all rental units
within one mile of the college campus, then it is natural to picture A by Figure 1,
called a Venn diagram.
The circle is an imaginary boundary that separates the elements of A (inside the
circle) from the elements of U that are not in A (outside the circle).
Figure 1 A is the shaded region
The Venn diagram of Figure 1 can be used to picture any set A and universal set
U. The universal set is the set of all elements under consideration. It is customary
to place the label A near the circle itself, but the elements of U are imagined to be
arranged so that the elements of U that are in A lie inside the circle and the elements
of U that are not in A lie outside the circle.
Let U be the set of all 26 lowercase letters of the English alphabet, let
A = 5a, b, c, d, e6, and let B = 5d, e, f, g, h, i6. Sets A and B are pictured in the
Venn diagram of Figure 2. Note that the elements d and e belong to both A and B
and that the elements a, b, c belong to A but not to B. Of course sets may have hun-
dreds or thousands of elements, so we seldom write the names of the elements as in
Figure 2; instead, we represent the situation by the Venn diagram of Figure 3 or by
the Venn diagram of Figure 4.
U U U
j k l m n o
A B A B
A B
a b d f g
z 3 2 4
c e h i y
p
q r x 17
s t u v w
Figure 2 Figure 3 Figure 4
The numbers in Figure 4 are interpreted as follows: There are 2 elements in the
overlap of A and B, 3 elements that belong to A but not B, 4 elements that belong to
B but not A, and 17 elements that belong to neither A nor B. Using such a diagram, it
is easy to calculate the number of elements in A (just add 2 and 3) or the number of
elements in B (just add 2 and 4).
The union of sets A and B, denoted by A ∪ B, is the set of elements formed by
combining all the elements of A and all the elements of B into one set.
DEFINITION Union
A ∪ B = 5x ∙ x ∊ A or x ∊ B6
Here we use the word or in the way it is always used in mathematics; that is, x
may be an element of set A or set B or both.
The union of two sets A and B is pictured in the Venn diagram of Figure 5. Note that
A B
A⊂A∪B and B⊂A∪B
The intersection of sets A and B, denoted by A ¨ B, is the set of elements in set
A that are also in set B.
Figure 5 A ∪ B is the shaded

region.
DEFINITION Intersection
A ¨ B = 5x ∙ x ∊ A and x ∊ B6
The intersection of two sets A and B is pictured in the Venn diagram of Figure 6.
A B Note that
A¨B⊂A and A¨B⊂B
If A ¨ B = ∅, then the sets A and B are said to be disjoint, as shown in Figure 7.
We now define an operation on sets called the complement. The complement of
Figure 6 A ¨ B is the shaded region. A (relative to a universal set U), denoted by A′, is the set of elements in U that are not
in A (Fig. 8).
A B
DEFINITION Complement
A = = 5x ∊ U ∙ x F A6
Figure 7 A ¨ B = ∅; A and B are U

disjoint.
A B
Figure 8 The shaded region is A′, the complement of A.
EXAMPLE 4 Union, Intersection, and Complement If A = 53, 6, 96, B = 53, 4, 5, 6, 76,

C = 54, 5, 76, and U = 51, 2, 3, 4, 5, 6, 7, 8, 96, then
A∪B = 53, 4, 5, 6, 7, 96
A¨B = 53, 66
A¨C = ∅ A and C are disjoint.
B′ = 51, 2, 8, 96
A B Matched Problem 4 If R = 51, 2, 3, 46, S = 51, 3, 5, 76, T = 52, 46,
and U = 51, 2, 3, 4, 5, 6, 7, 8, 96, find
n(A > B9) n(A > B) n(B > A9) (A) R ∪ S (B) R ¨ S (C) S ¨ T (D) S′
The number of elements in a set A is denoted by n1A2. So if A and B are

n(A9 > B9)
sets, then the numbers that are often shown in a Venn diagram, as in Figure 4, are
Figure 9 n1A ¨ B= 2, n1A ¨ B2, n1B ¨ A= 2, and n1A= ¨ B= 2 (see Fig. 9).
EXAMPLE 5 Numbers of Elements Let the universal set U be the set of positive integers less
than or equal to 100. Let A be the set of multiples of 3 in U, and let B be the set of
multiples of 5 in U.
(A) Find n1A ¨ B2, n1A ¨ B= 2, n1B ¨ A= 2, and n1A= ¨ B= 2.
(B) Draw a Venn diagram with circles labeled A and B, indicating the numbers of
elements in the subsets of part (A).
SOLUTION
(A) A = 53, 6, 9, c , 996, so n1A2 = 33.
B = 55, 10, 15, c , 1006, so n1B2 = 20.
A ¨ B = 515, 30, 45, c , 906, so n1A ¨ B2 = 6.
n1A ¨ B= 2 = 33 - 6 = 27
n1B ¨ A= 2 = 20 - 6 = 14
n1A= ¨ B= 2 = 100 - 16 + 27 + 142 = 53
(B)
A B
27 6 14
53
Matched Problem 5 Let the universal set U be the set of positive integers less
than or equal to 100. Let A be the set of multiples of 3 in U, and let B be the set of
multiples of 7 in U.
(A) Find n1A ¨ B2, n1A ¨ B= 2, n1B ¨ A= 2, and n1A= ¨ B= 2.
(B) Draw a Venn diagram with circles labeled A and B, indicating the numbers of
Application
EXAMPLE 6 Exit Polling In the 2016 presidential election, an exit poll of 100 voters produced
the results in the table (23 men voted for Clinton, 29 women for Clinton, 26 men for
Trump, and 22 women for Trump).
Men Women
Clinton 23 29
Trump 26 22
Let the universal set U be the set of 100 voters, C the set of voters for Clinton,
T the set of voters for Trump, M the set of male voters, and W the set of female
voters.
(A) Find n1C ¨ M2, n1C ¨ M = 2, n1M ¨ C = 2, and n1C = ¨ M = 2.
(B) Draw a Venn diagram with circles labeled C and M, indicating the numbers of
SOLUTION
(A) The set C = is equal to the set T, and the set M = is equal to the set W.
n1C ¨ M2 = 23, n1C ¨ M = 2 = 29
n1M ¨ C = 2 = 26, n1C = ¨ M = 2 = 22
(B)
C M
29 23 26
22
Matched Problem 6 Refer to Example 6.

(A) Find n1T ¨ W2, n1T ¨ W = 2, n1W ¨ T = 2, and n1T = ¨ W = 2.
(B) Draw a Venn diagram with circles labeled T and W, indicating the numbers of
In Example 6, find the number of voters in the set 1C ∪ M2 ¨ M′. Describe this set
verbally and with a Venn diagram.
EXAMPLE 7 Insurance Using a random sample of 100 insurance customers, an insurance

company generated the Venn diagram in Figure 10 where A is the set of cus-
tomers who purchased auto insurance, H is the set of customers who purchased
homeowner’s insurance, and L is the set of customers who purchased life insurance.
(A) How many customers purchased auto insurance?
(B) Shade the region H ∪ L in Figure 10. Find n1H ∪ L2.
A H
(C) Shade the region A ¨ H ¨ L′ in Figure 10. Find n1A ¨ H ¨ L′2.
16
50 15 SOLUTION
(A) The number of customers who purchased auto insurance is the number of cus-
8
2 6 tomers in the set A. We compute n(A) by adding all of the numbers in the Venn
diagram that lie inside the circle labeled A. This gives
1
L 50 + 16 + 8 + 2 = 76.
2
(B) We compute n1H ∪ L2 by adding all of the numbers in Figure 10
Figure 10 that lie inside the circle labeled H or the circle labeled L; that is,
n1H ∪ L2 = 15 + 16 + 8 + 6 + 2 + 1 = 48. Out of the 100 surveyed cus-
tomers, 48 purchased homeowner’s insurance or life insurance.
A H A H
16 16
50 15 50 15
8 8
2 6 2 6
1 1
L L
2 2
HøL A ù H ù L9
(C) Using Figure 10, we get that n1A ¨ H ¨ L′2 = 16. Out of the 100 surveyed
customers, there were 16 who purchased auto insurance and homeowner’s in-
surance and did not purchase life insurance.

(A) How many customers purchased homeowner’s insurance?
(B) Shade the region A ∪ L in Figure 10. Find n1A ∪ L2.
(C) Shade the region A′ ¨ H′ ¨ L in Figure 10. Find n1A′ ¨ H′ ¨ L2.
! CAUTION Note that in part (B), n1H2 = 45 and n1L2 = 17, but
n1H ∪ L2 ∙ 45 + 17 = 62. We do not count the customers in
the intersection of H and L twice.
Exercises 6.2
Skills Warm-up Exercises 4. Is the set of integers the union of the set of even integers and
the set of odd integers?
W In Problems 1–6, answer yes or no. (If necessary, review
Section B.1). 5. If the universal set is the set of integers, is the set of
positive integers the complement of the set of negative
1. Is the set of even integers a subset of the set of odd integers? integers?
2. Is the set of rational numbers a subset of the set of integers? 6. If the universal set is the set of integers, is the set of even
3. Is the set of integers the intersection of the set of even integers the complement of the set of odd integers?
integers and the set of odd integers?
A In Problems 7–14, indicate true (T) or false (F). C In Problems 49–52, determine whether the given set is finite or in-
finite. Consider the set N of positive integers to be the universal set.
7. 51, 26 ⊂ 52, 16 8. 53, 2, 16 ⊂ 51, 2, 3, 46
49. 5n ∈ N ∙ n 7 1006 50. 5n ∈ N ∙ n 6 10006
9. 55, 106 = 510, 56 10. 1 ∊ 510, 116
51. 52, 3, 5, 7, 11, 136 52. 52, 4, 6, 8, 10, c6
11. 506 ∊ 50, 5066 12. 50, 66 = 566
13. 8 ∊ 51, 2, 46 14. ∅ ⊂ 51, 2, 36 In Problems 53–58, draw a Venn diagram for sets A, B, and C and
shade the given region.
In Problems 15–28, write the resulting set using the listing 53. A ¨ B′ ¨ C 54. A′ ¨ B′ ¨ C
method.
55. 1A ¨ B2 ′ 56. 1A ∪ B2 ′
15. 51, 2, 36 ¨ 52, 3, 46 16. 52, 3, 56 ¨ 52, 4, 66
57. A′ ∪ 1B′ ¨ C2 58. 1A ¨ B2 ′ ∪ C
17. 51, 2, 36 ∪ 52, 3, 46 18. 52, 3, 56 ∪ 52, 4, 66
19. 51, 4, 76 ∪ 510, 136 20. 5 - 3, - 16 ¨ 51, 36 In Problems 59–62, are the given sets disjoint? Let H, T, P, and E
denote the sets in Problems 49, 50, 51, and 52, respectively.
21. 51, 4, 76 ¨ 510, 136 22. 5 - 3, - 1,6 ∪ 51, 36
59. H′ and T′ 60. T′ and P
2
B 23. 5x ∙ x = 256 24. 5x ∙ x2 = 366
61. P′ and H 62. E′ and T
25. 5x ∙ x3 = - 276 26. 5x ∙ x4 = 2566
In Problems 63–72, discuss the validity of each statement. Venn
27. 5x ∙ x is an odd number between 1 and 9, inclusive6
diagrams may be helpful. If the statement is true, explain why. If
28. 5x ∙ x is a month starting with M6 not, give a counterexample.
29. For U = 51, 2, 3, 4, 56 and A = 52, 3, 46, find A′. 63. If A ⊂ B, then A ¨ B = A.
30. For U = 52, 3, 5, 7, 11, 136 and B = 53, 136, find B′. 64. If A ⊂ B, then A ∪ B = A.
65. If A ∪ B = A, then A ⊂ B.
In Problems 31–44, refer to the Venn diagram below and find the
indicated number of elements. 66. If A ¨ B = A, then A ⊂ B.
U 67. If A ¨ B = ∅, then A = ∅.
A B 68. If A = ∅, then A ¨ B = ∅.
69. If A ⊂ B, then A′ ⊂ B′.
21 42 18
70. If A ⊂ B, then B′ ⊂ A′.
19 71. The empty set is an element of every set.
72. The empty set is a subset of the empty set.
31. n1U2 32. n1A2
73. How many subsets does each of the following sets contain?
33. n1B2 34. n1A ¨ B2
(A) 5a6
35. n1A ∪ B2 36. n1B′2
(B) 5a, b6
37. n1A′2 38. n1A ¨ B′2
(C) 5a, b, c6
39. n1B ¨ A′2 40. n1 1A ¨ B2 ′2
(D) 5a, b, c, d6
41. n1 1A ∪ B2 ′2 42. n1A′ ¨ B′2
74. Let A be a set that contains exactly n elements. Find a for-
43. n1A ∪ B′2 44. n1A′ ∪ B2 mula in terms of n for the number of subsets of A.
45. If R = 51, 2, 3, 46 and T = 52, 4, 66, find
(A) 5x ∙ x ∊ R or x ∊ T6
(B) R ∪ T Applications
46. If R = 51, 3, 46 and T = 52, 4, 66, find
Enrollments. In Problems 75–88, find the indicated number of
(A) 5x ∙ x ∊ R and x ∊ T6 elements by referring to the following table of enrollments in a
finite mathematics class:
(B) R ¨ T
47. For P = 51, 3, 56, Q = 52, 3, 56, and R = 52, 4, 6, 86, Freshmen Sophomores
find P ¨ 1Q ∪ R2. Arts & Sciences 19 14
48. For P, Q, and R in Problem 47, find P ∪ 1Q ¨ R2. Business 66 21

Let the universal set U be the set of all 120 students in the class, A 95. 1A ∪ B2 ′ 96. 1A ∪ B ∪ Rh2 ′
the set of students from the College of Arts & Sciences, B the set of
97. A′ ¨ B 98. Rh′ ¨ A
students from the College of Business, F the set of freshmen, and S
the set of sophomores.
75. n1F2 76. n1S2 1. (A) 5x ∙ x2 = 96 (B) 5 - 3, 36 (C) True; true
77. n1A2 78. n1B2 2. All are true.
3. 51, 26, 516, 526, ∅
79. n1A ¨ S2 80. n1A ¨ F2 4. (A) 51, 2, 3, 4, 5, 76 (B) 51, 36
81. n1B ¨ F2 82. n1B ¨ S2 (C) ∅ (D) 52, 4, 6, 8, 96
5. (A) n1A ¨ B2 = 4; n1A ¨ B′2 = 29; n1B ¨ A′2 = 10;
83. n1A ∪ S2 84. n1A ∪ F2 n1A′ ¨ B′2 = 57
85. n1B ∪ F2 86. n1B ∪ S2 (B)
A B
87. n1A ¨ B2 88. n1F ∪ S2
89. Committee selection. A company president and three vice- 29 4 10
presidents, denoted by the set 5P, V1, V2, V3 6, wish to select
a committee of 2 people from among themselves. How many 57
ways can this committee be formed? That is, how many
2-person subsets can be formed from a set of 4 people? 6. (A) n1T ¨ W2 = 22; n1T ¨ W′2 = 26; n1W ¨ T′2 = 29;
n1T′ ¨ W′2 = 23
90. Voting coalition. The company’s leaders in Problem 89 de- (B)
cide for or against certain measures as follows: The president T W
has 2 votes and each vice-president has 1 vote. Three favor-
able votes are needed to pass a measure. List all minimal 26 22 29
winning coalitions; that is, list all subsets of 5P, V1, V2, V3 6
that represent exactly 3 votes.
23
Blood types. When receiving a blood transfusion, a recipient
7. (A) n1H2 = 45
must have all the antigens of the donor. A person may have one or
more of the three antigens A, B, and Rh or none at all. Eight blood (B)
A H
types are possible, as indicated in the following Venn diagram,
where U is the set of all people under consideration: 16
50 15
An A - person has A antigens but no B or Rh antigens, an O+
8
person has Rh but neither A nor B, an AB - person has A and B 2 6
but no Rh, and so on.
U
1
A B L
Blood Types 2
AB2 A2 n(A ø L) 5 83
A2 B2
A1
AB1 B2
(C)
A H
A1 B1 B1
AB2 16
AB1 50 15
O1 O2
O2
Rh O1 8
2 6
In Problems 91–98, use the Venn diagram to indicate which of the
eight blood types are included in each set. 1
L
91. A ¨ Rh 92. A ¨ B 2
n(A9 ù H9 ù L) 5 1
93. A ∪ Rh 94. A ∪ B
6.3 Basic Counting Principles

■■ Addition Principle Addition Principle
■■ Multiplication Principle If the enrollment in a college chemistry class consists of 13 males and 15 females,
then there are a total of 28 students enrolled in the class. This is a simple example
of a counting technique, a method for determining the number of elements in a set
SECTION 6.3 Basic Counting Principles 347
without actually enumerating the elements one by one. Set operations play a funda-
mental role in many counting techniques. For example, if M is the set of male students
in the chemistry class and F is the set of female students, then the union of sets M and
F, denoted M ∪ F, is the set of all students in the class. Since these sets have no ele-
ments in common, the intersection of sets M and F, denoted M ¨ F, is the empty set
∅; we then say that M and F are disjoint sets. The total number of students enrolled
in the class is the number of elements in M ∪ F, denoted by n1M ∪ F2 and given by
n1M ∪ F2 = n1M2 + n1F2
= 13 + 15 = 28
In this example, the number of elements in the union of sets M and F is the sum
of the number of elements in M and in F. However, this does not work for all pairs
of sets. To see why, consider another example. Suppose that the enrollment in a
mathematics class consists of 22 math majors and 16 physics majors, and that 7 of
these students have majors in both subjects. If M represents the set of math majors
and P represents the set of physics majors, then M ¨ P represents the set of double
majors. It is tempting to proceed as before and conclude that there are 22 + 16 = 38
students in the class, but this is incorrect. We have counted the double majors twice,
once as math majors and again as physics majors. To correct for this double counting,
we subtract the number of double majors from this sum. Thus, the total number of
students enrolled in this class is given by
n1M ∪ P2 = n1M2 + n1P2 - n1M ¨ P2 (1)
= 22 + 16 - 7 = 31
Equation (1) illustrates the addition principle for counting the elements in the
union of two sets.
THEOREM 1 Addition Principle (for Counting)

For any two sets A and B,
n1A ∪ B2 = n1A2 + n1B2 - n1A ¨ B2 (2)
Note that if A and B are disjoint, then n1A ¨ B2 = 0, and equation (2) becomes
n1A ∪ B2 = n1A2 + n1B2.
EXAMPLE 1 Employee Benefits According to a survey of business firms in a certain city,

750 firms offer their employees health insurance, 640 offer dental insurance, and
280 offer health insurance and dental insurance. How many firms offer their em-
ployees health insurance or dental insurance?
SOLUTION If H is the set of firms that offer their employees health insurance and
D is the set that offer dental insurance, then
H ¨ D = set of firms that offer health insurance and dental insurance
H ∪ D = set of firms that offer health insurance or dental insurance
So
n1H2 = 750 n1D2 = 640 n1H ¨ D2 = 280
and
n1H ∪ D2 = n1H2 + n1D2 - n1H ¨ D2
= 750 + 640 - 280 = 1,110
Therefore, 1,110 firms offer their employees health insurance or dental insurance.
Matched Problem 1 The survey in Example 1 also indicated that 345 firms
offer their employees group life insurance, 285 offer long-term disability insur-
ance, and 115 offer group life insurance and long-term disability insurance. How
many firms offer their employees group life insurance or long-term disability
insurance?
EXAMPLE 2 Market Research A survey of 100 families found that 35 families subscribe to
the video streaming service Webfilms, 60 families subscribe to the video streaming
service Nile Prime, and 20 families subscribe to both video streaming services.
(A) How many families subscribe to Webfilms but not Nile Prime?
(B) How many subscribe to Nile Prime but not Webfilms?
(C) How many do not subscribe to either video streaming service?
U (D) Organize this information in a table.
W N
SOLUTION Let U be the group of families surveyed. Let W be the set of families
that subscribe to Webfilms, and let N be the set of families that subscribe to Nile
W > N9 W > N W9 > N Prime. Since U contains all the elements under consideration, it is the universal set
for this problem. The complement of W, denoted by W′, is the set of families in the
survey group U that do not subscribe to Webfilms. Similarly, N′ is the set of fami-
W9 > N9
lies in the group that do not subscribe to Nile Prime. Using the sets W and N, their
Figure 1 Venn diagram for the video complements, and set intersection, we can divide U into the four disjoint subsets
streaming survey defined below and illustrated as a Venn diagram in Figure 1.
W¨N = set of families that subscribe to both services
W ¨ N′ = set of families that subscribe to Webfilms but not Nile Prime
W′ ¨ N = set of families that subscribe to Nile Prime but not Webfilms
W′ ¨ N′ = set of families that do not subscribe to either service
The given survey information can be expressed in terms of set notation as
n1U2 = 100 n1W2 = 35 n1N2 = 60 n1W ¨ N2 = 20
U We can use this information and a Venn diagram to answer parts (A)–(C). First, we
W N place 20 in W ¨ N in the diagram (see Fig. 2). As we proceed with parts (A) through
(C), we add each answer to the diagram.
15 20 40
(A) Since 35 families subscribe to Webfilms and 20 subscribe to both services, the
number of families that subscribe to Webfilms but not to Nile Prime is
25
N1W ¨ N′2 = 35 - 20 = 15.
Figure 2 Survey results
(B) In a similar manner, the number of families that subscribe to Nile Prime but
not Webfilms is
N1W′ ¨ N2 = 60 - 20 = 40.
(C) The total number of subscribers is 20 + 15 + 40 = 75. The total number of
families that do not subscribe to either video streaming service is
n1W′ ¨ N′2 = 100 - 75 = 25.
(D) Venn diagrams are useful tools for determining the number of elements in the
various sets of a survey, but often the results must be presented in the form of
a table, rather than a diagram. Table 1 contains the information in Figure 2 and
also includes totals that give the numbers of elements in the sets W, W′, N, N′,
and U.
Table 1
Nile Prime
Subscriber, N Nonsubscriber, N′ Totals
Subscriber, W 20 15 35
Webfilms Nonsubscriber, W′ 40 25 65
Totals 60 40 100
CONCEPTUAL INSIGHT
Carefully compare Table 1 and Figure 2 of Example 2, and note that we do not
include any of the totals in Table 1 in the Venn diagram. Instead, the numbers
in the Venn diagram give the numbers of elements in the four disjoint sets
W ¨ N, W ¨ N′, W′ ¨ N, and W′ ¨ N′. From Figure 2, it is easy to construct
Table 1 (the totals are easily calculated from the Venn diagram). And from
Table 1, it is easy to construct the Venn diagram of Figure 2 (simply disregard
the totals).
Matched Problem 2 Students at a university have the option to stream TV

shows over the internet or to watch cable TV. A survey of 100 college students
produced the following results: In the past 30 days, 65 people have streamed TV
shows, 45 have watched cable TV, and 30 have done both.
(A) During this 30-day period, how many people in the survey have streamed TV
shows but not watched cable TV?
(B) How many have watched cable TV but not streamed TV shows?
(C) How many have done neither?
(D) Organize this information in a table.
U
A B Explore and Discuss 1
Let A, B, and C be three sets. Use a Venn diagram (Fig. 3) to explain the following
equation:
n1A ∪ B ∪ C2 = n1A2 + n1B2 + n1C2 - n1A ¨ B2 - n1A ¨ C2
- n1B ¨ C2 + n1A ¨ B ¨ C2
C
Figure 3 Multiplication Principle

As we have just seen, if the elements of a set are determined by the union operation,
addition and subtraction are used to count the number of elements in the set. Now we
want to consider sets whose elements are determined by a sequence of operations.
We will see that multiplication is used to count the number of elements in sets formed
this way.
EXAMPLE 3 Product Mix A retail store stocks windbreaker jackets in small, medium, large,
and extra large. All are available in blue or red. What are the combined choices, and
how many combined choices are there?
SOLUTION To solve the problem, we use a tree diagram:
SIZE CHOICES COLOR CHOICES COMBINED CHOICES

(OUTCOMES) (OUTCOMES) (OUTCOMES)
B (S, B)
S
R (S, R)
B (M, B)
M
R (M, R)
Start
B (L, B)
L
R (L, R)
B (XL, B)
XL
R (XL, R)
There are 8 possible combined choices (outcomes). There are 4 ways that a size can
be chosen and 2 ways that a color can be chosen. The first element in the ordered
pair represents a size choice, and the second element represents a color choice.
Matched Problem 3 A company offers its employees health plans from three
different companies: R, S, and T. Each company offers two levels of coverage, A
and B, with one level requiring additional employee contributions. What are the
combined choices, and how many choices are there? Solve using a tree diagram.
Suppose that you asked, “From the 26 letters in the alphabet, how many ways
can 3 letters appear on a license plate if no letter is repeated?” To try to count the pos-
sibilities using a tree diagram would be extremely tedious. The following multiplica-
tion principle will enable us to solve this problem easily. In addition, it forms the
basis for developing other counting devices in the next section.
THEOREM 2 Multiplication Principle (for Counting)

1. If two operations O1 and O2 are performed in order, with N1 possible outcomes
for the first operation and N2 possible outcomes for the second operation, then
there are
N1 # N2
possible combined outcomes of the first operation followed by the second.
2. In general, if n operations O1, O2, c, On are performed in order, with possible
number of outcomes N1, N2, c, Nn, respectively, then there are
N1 # N2 # g # Nn
possible combined outcomes of the operations performed in the given order.
In Example 3, we see that there are 4 possible outcomes in choosing a size

(the first operation) and 2 possible outcomes in choosing a color (the second
operation). So by the multiplication principle, there are 4 # 2 = 8 possible com-
bined outcomes. Use the multiplication principle to solve Matched Problem 3.
[Answer: 3 # 2 = 6]
To answer the license plate question: There are 26 ways the first letter can be
chosen; after a first letter is chosen, there are 25 ways a second letter can be chosen;
and after 2 letters are chosen, there are 24 ways a third letter can be chosen. So, using
the multiplication principle, there are 26 # 25 # 24 = 15,600 possible 3-letter license
plates if no letter is repeated.
EXAMPLE 4 Computer-Assisted Testing Many colleges and universities use computer-assist-

ed testing. Suppose that a screening test is to consist of 5 questions, and a computer
stores 5 comparable questions for the first test question, 8 for the second, 6 for the
third, 5 for the fourth, and 10 for the fifth. How many different 5-question tests can
the computer select? (Two tests are considered different if they differ in one or more
questions.)
SOLUTION O1: Selecting the first question N1: 5 ways
O2: Selecting the second question N2: 8 ways
O3: Selecting the third question N3: 6 ways
O4: Selecting the fourth question N4: 5 ways
O5: Selecting the fifth question N5: 10 ways
The computer can generate

5 # 8 # 6 # 5 # 10 = 12,000 different tests
Matched Problem 4 Each question on a multiple-choice test has 5 choices.

If there are 5 such questions on a test, how many different responses are possible if
only 1 choice is marked for each question?
EXAMPLE 5 Code Words How many 3-letter code words are possible using the first 8 letters
of the alphabet if
(A) No letter can be repeated?
(B) Letters can be repeated?
(C) Adjacent letters cannot be alike?
SOLUTION To form 3-letter code words from the 8 letters available, we select a let-
ter for the first position, one for the second position, and one for the third position.
Altogether, there are three operations.
(A) No letter can be repeated:
O1: Selecting the first letter N1: 8 ways
O2: Selecting the second letter N2: 7 ways Since 1 letter has been used
O3: Selecting the third letter N3: 6 ways Since 2 letters have been used
There are
8 # 7 # 6 = 336 possible code words Possible combined operations
(B) Letters can be repeated:
O2: Selecting the second letter N2: 8 ways Repeats allowed
O3: Selecting the third letter N3: 8 ways Repeats allowed
There are
8 # 8 # 8 = 83 = 512 possible code words
(C) Adjacent letters cannot be alike:
O2: Selecting the second letter N2: 7 ways Cannot be the same as the first
O3: Selecting the third letter N3: 7 ways Cannot be the same as the
second, but can be the same
as the first
There are
8 # 7 # 7 = 392 possible code words
Matched Problem 5 How many 4-letter code words are possible using the first
10 letters of the alphabet under the three different conditions stated in Example 5?
Exercises 6.3
Skills Warm-up Exercises (A) If a golfer decides to play a round at a silver or gold
course, how many selections are possible?
W In Problems 1–6, Solve for x. (If necessary, review Section A.1).
(B) If a golfer decides to play one round per week for 3
1. 50 = 34 + 29 - x 2. 124 = 73 + 87 - x weeks, first on a bronze course, then silver, then gold,
3. 4x = 23 + 42 - x 4. 7x = 51 + 45 - x how many combined selections are possible?
5. 31x + 112 = 65 + x - 14 16. The 14 colleges of interest to a high school senior include 6
that are expensive (tuition more than $30,000 per year), 7 that
6. 111x + 92 = 234 + x - 65 are far from home (more than 200 miles away), and 2 that are
both expensive and far from home.
A Solve Problems 7–10 two ways: (A) using a tree diagram, and (B)
using the multiplication principle. (A) If the student decides to select a college that is not
expensive and within 200 miles of home, how many
7. How many ways can 2 coins turn up—heads, H, or tails, selections are possible?
T—if the combined outcome (H, T) is to be distinguished
from the outcome (T, H)? (B) If the student decides to attend a college that is not
expensive and within 200 miles from home during his
8. How many 2-letter code words can be formed from the first first two years of college, and then will transfer to a
3 letters of the alphabet if no letter can be used more than once? college that is not expensive but is far from home, how
9. A coin is tossed with possible outcomes of heads H, or tails many selections of two colleges are possible?
T. Then a single die is tossed with possible outcomes 1, 2, 3, In Problems 17–24, use the given information to determine the
4, 5, or 6. How many combined outcomes are there? number of elements in each of the four disjoint subsets in the fol-
10. In how many ways can 3 coins turn up—heads H, or tails lowing Venn diagram.
T—if combined outcomes such as (H, T, H), (H, H, T), and U
(T, H, H) are considered as different? A B
11. A London entertainment guide recommends 12 restaurants

A > B9 A > B A9 > B
and 7 musicals that appeal to a couple.
(A) If the couple goes to dinner or a musical, but not both,
how many selections are possible? A9 > B9
(B) If the couple goes to dinner and then to a musical, how 17. n1A2 = 100, n1B2 = 90, n1A ¨ B2 = 50, n1U2 = 200
many combined selections are possible?
18. n1A2 = 40, n1B2 = 60, n1A ¨ B2 = 20, n1U2 = 100
12. A college offers 2 introductory courses in history, 3 in sci-
ence, 2 in mathematics, 2 in philosophy, and 1 in English. 19. n1A2 = 35, n1B2 = 85, n1A ∪ B2 = 90, n1U2 = 100
(A) If a freshman takes one course in each area during her 20. n1A2 = 65, n1B2 = 150, n1A ∪ B2 = 175, n1U2 = 200
first semester, how many course selections are possible?
21. n1A′2 = 110, n1B′2 = 220, n1A′ ¨ B′2 = 60, n1U2 = 300
(B) If a part-time student can afford to take only one intro-
ductory course, how many selections are possible? 22. n1A′2 = 40, n1B′2 = 70, n1A′ ¨ B′2 = 10, n1U2 = 500
13. How many 3-letter code words can be formed from the 23. n1A′2 = 20, n1B′2 = 40, n1A′ ∪ B′2 = 50, n1U2 = 80
letters A, B, C, D, E if no letter is repeated? If letters can be 24. n1A′2 = 60, n1B′2 = 20, n1A′ ∪ B′2 = 75, n1U2 = 250
repeated? If adjacent letters must be different?
14. How many 4-letter code words can be formed from the letters In Problems 25–32, use the given information to complete the
A, B, C, D, E, F, G if no letter is repeated? If letters can be following table.
repeated? If adjacent letters must be different? A A′ Totals
B 15. A county park system rates its 20 golf courses in increasing B ? ? ?
order of difficulty as bronze, silver, or gold. There are only B∙ ? ? ?
two gold courses and twice as many bronze as silver courses. Totals ? ? ?
25. n1A2 = 70, n1B2 = 90, 42. How many 5-digit ZIP code numbers are possible? How
n1A ¨ B2 = 30, n1U2 = 200 many of these numbers contain no repeated digits?
26. n1A2 = 55, n1B2 = 65, 43. In Example 3, does it make any difference in which order
the selection operations are performed? That is, if we select
n1A ¨ B2 = 35, n1U2 = 100 a jacket color first and then select a size, are there as many
27. n1A2 = 35, n1B2 = 75, combined choices available as selecting a size first and then
a color? Justify your answer using tree diagrams and the
n1A ∪ B2 = 80, n1U2 = 100 multiplication principle.
28. n1A2 = 80, n1B2 = 70, 44. Explain how three sets, A, B, and C, can be related to each
n1A ∪ B2 = 110, n1U2 = 200 other in order for the following equation to hold true (Venn
diagrams may be helpful):
29. n1A′2 = 25, n1B′2 = 64,
n1A ∪ B ∪ C2 = n1A2 + n1B2 + n1C2
n1A′ ∪ B′2 = 81, n1U2 = 120
- n1A ¨ C2 - n1B ¨ C2
30. n1A′2 = 81, n1B′2 = 90,
C Problems 45–48 refer to the following Venn diagram.
n1A′ ¨ B′2 = 63, n1U2 = 180
U
31. n1A2 = 110, n1B2 = 145, A B
n1A ∪ B2 = 255, n1U2 = 300

x y z
32. n1A2 = 175, n1B2 = 125,
n1A ∪ B2 = 300, n1U2 = 300 w
In Problems 33 and 34, discuss the validity of each statement. If the 45. Which of the numbers x, y, z, or w must equal 0 if B ⊂ A?
statement is always true, explain why. If not, give a counterexample. 46. Which of the numbers x, y, z, or w must equal 0 if A and B
33. (A) If A or B is the empty set, then A and B are disjoint. are disjoint?
(B) If A and B are disjoint, then A or B is the empty set. 47. Which of the numbers x, y, z, or w must equal 0 if
A ∪ B = A ¨ B?
34. (A) If A and B are disjoint, then n1A ¨ B2 = n1A2 + n1B2.
48. Which of the numbers x, y, z, or w must equal 0 if
(B) If n1A ∪ B2 = n1A2 + n1B2, then A and B are disjoint. A ∪ B = U?
35. A particular new car model is available with 5 choices of 49. A group of 75 people includes 32 who play tennis, 37 who
color, 3 choices of transmission, 4 types of interior, and 2 play golf, and 8 who play both tennis and golf. How many
types of engine. How many different variations of this model people in the group play neither sport?
are possible?
50. A class of 30 music students includes 13 who play the piano,
36. A delicatessen serves meat sandwiches with the following 16 who play the guitar, and 5 who play both the piano and the
options: 3 kinds of bread, 5 kinds of meat, and lettuce or guitar. How many students in the class play neither instrument?
sprouts. How many different sandwiches are possible, assum-
ing that one item is used out of each category? 51. A group of 100 people touring Europe includes 42 people
who speak French, 55 who speak German, and 17 who speak
37. Using the English alphabet, how many 6-character case- neither language. How many people in the group speak both
sensitive passwords are possible? French and German?
38. Using the English alphabet, how many 6-character case- 52. A high school football team with 40 players includes 16 play-
sensitive passwords are possible if each character is a letter ers who played offense last year, 17 who played defense, and
or a digit? 12 who were not on last year’s team. How many players from
39. A combination lock has 5 wheels, each labeled with the 10 last year played both offense and defense?
digits from 0 to 9. How many 5-digit opening combinations
are possible if no digit is repeated? If digits can be repeated?
If successive digits must be different?
40. A small combination lock has 3 wheels, each labeled with
Applications
the 10 digits from 0 to 9. How many 3-digit combinations are
53. Management. A management selection service classifies its
possible if no digit is repeated? If digits can be repeated? If
applicants (using tests and interviews) as high-IQ, middle-IQ,
successive digits must be different?
or low-IQ and as aggressive or passive. How many combined
41. How many different license plates are possible if each classifications are possible?
contains 3 letters (out of the alphabet’s 26 letters) followed
(A) Solve using a tree diagram.
by 3 digits (from 0 to 9)? How many of these license plates
contain no repeated letters and no repeated digits? (B) Solve using the multiplication principle.
54. Management. A corporation plans to fill 2 different posi- Workers per Age Group
tions for vice-president, V1 and V2, from administrative (thousands)
officers in 2 of its manufacturing plants. Plant A has 6 of- 16–19 20–24 25∙ Totals
ficers and plant B has 8. How many ways can these 2 posi-
Males at 343 154 237 734
tions be filled if the V1 position is to be filled from plant A
Minimum Wage
and the V2 position from plant B? How many ways can the
2 positions be filled if the selection is made without regard Males below 118 102 159 379
Minimum Wage
to plant?
Females at 367 186 503 1,056
55. Transportation. A sales representative who lives in city A Minimum Wage
wishes to start from home and visit 3 different cities: B, C, Females below 251 202 540 993
and D. She must choose whether to drive her own car or to Minimum Wage
fly. If all cities are interconnected by both roads and airlines, Totals 1,079 644 1,439 3,162
how many travel plans can be constructed to visit each city
exactly once and return home?
(B) How many females are of age 20 or older and earn
56. Transportation. A manufacturing company in city A minimum wage?
wishes to truck its product to 4 different cities: B, C, D,
and E. If roads interconnect all 4 cities, how many differ- (C) How many workers are of age 16–19 or males earning
ent route plans can be constructed so that a single truck, minimum wage?
starting from A, will visit each city exactly once, then
(D) How many workers earn below minimum wage?
return home?
62. Minimum wage. Refer to the table in Problem 61.
57. Market research. A survey of 1,500 people indicates
that 565 own dogs, 372 own cats, and 128 own dogs and (A) How many females are of age 16–19 and earn minimum
cats. wage?
(A) How many people in the survey own either a dog or (B) How many males are of age 16–24 and earn below
a cat? minimum wage?
(B) How many own neither a dog nor a cat? (C) How many workers are of age 20–24 or females earning
below minimum wage?
(C) How many own a dog and do not own a cat?
(D) How many workers earn minimum wage?
58. Market research. A survey of 600 luxury hotels indicates
that 230 own a gym, 270 own a spa, and 120 own a gym 63. Medicine. A medical researcher classifies subjects according
and a spa. to male or female; smoker or nonsmoker; and underweight,
average weight, or overweight. How many combined clas-
(A) How many luxury hotels in the survey own either a gym
sifications are possible?
or a spa?
(A) Solve using a tree diagram.
(B) How many own neither a gym nor a spa?
(B) Solve using the multiplication principle.
(C) How many own a spa and do not own a gym?
64. Family planning. A couple is planning to have 3 children.
59. Communications. A cable television company has 8,000
How many boy–girl combinations are possible? Distinguish
subscribers in a suburban community. The company offers
between combined outcomes such as (B, B, G), (B, G, B), and
two premium channels: HBO and Showtime. If 2,450
(G, B, B).
subscribers receive HBO, 1,940 receive Showtime, and 5,180
do not receive any premium channel, how many subscribers (A) Solve using a tree diagram.
receive both HBO and Showtime?
(B) Solve using the multiplication principle.
60. Communications. A cable company offers its 10,000 cus-
65. Politics. A politician running for a third term is planning to
tomers two special services: high-speed internet and digital
contact all contributors to her first two campaigns. If 1,475
phone. If 3,770 customers use high-speed internet, 3,250 use
individuals contributed to the first campaign, 2,350 contrib-
digital phone, and 4,530 do not use either of these services,
uted to the second campaign, and 920 contributed to the first
how many customers use both high-speed internet and digital
and second campaigns, how many individuals have contrib-
phone?
uted to the first or second campaign?
61. Minimum wage. The table gives the number of male and
66. Politics. If 12,457 people voted for a politician in his first
female workers earning at or below the minimum wage for
election, 15,322 voted for him in his second election, and
several age categories.
9,345 voted for him in the first and second elections, how
(A) How many males are of age 20–24 and earn below mini- many people voted for this politician in the first or second
mum wage? election?
SECTION 6.4 Permutations and Combinations 355
Answers to Matched Problems 3. There are 6 combined choices:

COMPANY COVERAGE COMBINED
1. 515
CHOICES CHOICES CHOICES
2. (A) 35
(OUTCOMES) (OUTCOMES) (OUTCOMES)
(B) 15
(C) 20 A (R, A)
R
(D) B (R, B)
Cable TV A (S, A)
Start S
Viewer Nonviewer Totals B (S, B)
Streamer 30 35 65
A (T, A)
T
B (T, B)
5. (A) 10 # 9 # 8 # 7 = 5,040
Internet Nonstreamer 15 20 35
4. 55, or 3,125
Totals 45 55 100 (B) 10 # 10 # 10 # 10 = 10,000
(C) 10 # 9 # 9 # 9 = 7,290
6.4 Permutations and Combinations

■■ Factorials The multiplication principle discussed in the preceding section can be used to
■■ Permutations
develop two additional counting devices that are extremely useful in more compli-
cated counting problems. Both of these devices use factorials.
■■ Combinations
■■ Applications Factorials
When using the multiplication principle, we encountered expressions such as
26 # 25 # 24 or 8#7#6
where each natural number factor is decreased by 1 as we move from left to right. The
factors in the following product continue to decrease by 1 until a factor of 1 is reached:
5#4#3#2#1
Products like this are encountered so frequently in counting problems that it is use-
ful to express them in a concise notation. The product of the first n natural numbers
is called n factorial and is denoted by n! Also, we define zero factorial, 0!, to be 1.
DEFINITION Factorial*
For a natural number n,
n! = n1n - 121n - 22 # g # 2 # 1 4! = 4 # 3 # 2 # 1
0! = 1
n! = n # 1n - 12!
EXAMPLE 1 Computing Factorials

(A) 5! = 5 # 4 # 3 # 2 # 1 = 120
7! 7 # 6!
(B) = = 7
6! 6!
8 # 7 # 6 # 5!
= 8 # 7 # 6 = 336
8!
(C) =
5! 5!
52! 52 # 51 # 50 # 49 # 48 # 47!
5 # 4 # 3 # 2 # 1 # 47!
(D) = = 2,598,960
5!47!
*
Many calculators have an n! key or its equivalent.
Matched Problem 1 Find

10! 10! 5! 20!
(A) 6! (B) (C) (D) (E)
9! 7! 0!3! 3!17!
It is interesting and useful to note that n! grows very rapidly. Compare the
following:
5! = 120 10! = 3,628,800 15! = 1,307,674,368,000
Try 69!, 70!, and 71! on your calculator.
Permutations
A particular (horizontal or vertical) arrangement of a set of paintings on a wall is
called a permutation of the set of paintings.
DEFINITION Permutation of a Set of Objects

A permutation of a set of distinct objects is an arrangement of the objects in a
specific order without repetition.
Suppose that 4 pictures are to be arranged from left to right on one wall of an
art gallery. How many permutations (ordered arrangements) are possible? Using the
multiplication principle, there are 4 ways of selecting the first picture; after the first
picture is selected, there are 3 ways of selecting the second picture. After the first 2
pictures are selected, there are 2 ways of selecting the third picture, and after the first
3 pictures are selected, there is only 1 way to select the fourth. So the number of per-
mutations (ordered arrangements) of the set of 4 pictures is
4 # 3 # 2 # 1 = 4! = 24
In general, how many permutations of a set of n distinct objects are possible?
Reasoning as above, there are n ways in which the first object can be chosen, there
are n - 1 ways in which the second object can be chosen, and so on. Using the mul-
tiplication principle, we have the following:
THEOREM 1 Number of Permutations of n Objects

The number of permutations of n distinct objects without repetition, denoted by
nPn, is
nPn = n1n - 12 # g # 2 # 1 = n! n factors

Example: The number of permutations of 7 objects is
7P7 = 7 # 6 # 5 # 4 # 3 # 2 # 1 = 7! 7 factors
Now suppose that the director of the art gallery decides to use only 2 of the 4
available paintings, and they will be arranged on the wall from left to right. We are
now talking about a particular arrangement of 2 paintings out of the 4, which is called
a permutation of 4 objects taken 2 at a time. In general,
DEFINITION Permutation of n Objects Taken r at a Time

A permutation of a set of n distinct objects taken r at a time without repetition is an
arrangement of r of the n objects in a specific order.
How many ordered arrangements of 2 pictures can be formed from the 4? That
is, how many permutations of 4 objects taken 2 at a time are there? There are 4 ways
that the first picture can be selected; after selecting the first picture, there are 3 ways
that the second picture can be selected. So the number of permutations of a set of 4
objects taken 2 at a time, which is denoted by 4P2, is given by
4P2 = 4#3
In terms of factorials, we have
4 # 3 # 2!
= 4#3 =
4!
4P2 = Multiplying 4 # 3 by 1 in the form 2!>2!
2! 2!
Reasoning in the same way as in the example, we find that the number of permuta-
tions of n distinct objects taken r at a time without repetition 10 … r … n2 is given by
nPr = n1n - 121n - 22 # g # 1n - r + 12 r factors
# #
9P6 = 919 - 1219 - 22 g 19 - 6 + 12 6 factors
= 9#8#7#6#5#4
Multiplying the right side of the equation for nPr by 1 in the form 1n - r2!> 1n - r2!,
we obtain a factorial form for nPr:
1n - r2!
nPr = n1n - 121n - 22 # g # 1n - r + 12
1n - r2!
But, since
n1n - 121n - 22 # g # 1n - r + 121n - r2! = n!
the expression above simplifies to
n!
nPr =
1n - r2!
We summarize these results in Theorem 2.
THEOREM 2 Number of Permutations of n Objects Taken r at a Time

The number of permutations of n distinct objects taken r at a time without repetition
is given by*
nPr = n1n - 121n - 22 # g # 1n - r + 12 r factors
#
5P2 = 5 4 factors
or
n! 5! 5!
nPr = 0 … r … n 5P2 = =
1n - r2! 15 - 22! 3!
n! n!
Note: nPn = = = n! permutations of n objects taken n at a time.
1n - n2! 0!
Remember, by definition, 0! = 1.
* In place of the symbol nPr, the symbols Prn, Pn,r , and P1n, r2 are often used.
EXAMPLE 2 Permutations Given the set 5A, B, C6, how many permutations are possible for
this set of 3 objects taken 2 at a time? Answer the question
(A) Using a tree diagram
(B) Using the multiplication principle
(C) Using the two formulas for nPr
SOLUTION
(A) Using a tree diagram:
FIRST SECOND
CHOICE CHOICE PERMUTATION
B AB
A
C AC
A BA
Start B
C BC
A CA
C
B CB
There are 6 permutations of 3 objects taken 2 at a time.
(B) Using the multiplication principle:
O1: Fill the first position N1: 3 ways
O2: Fill the second position N2: 2 ways
There are
3 # 2 = 6 permutations of 3 objects taken 2 at a time
(C) Using the two formulas for nPr:
2 factors
T
3#2#1
= 3#2 = 6
3!
3P2 or 3P2 = = = 6
13 - 22! 1
There are 6 permutations of 3 objects taken 2 at a time. Of course, all three
methods produce the same answer.
Matched Problem 2 Given the set 5A, B, C, D6, how many permutations are
possible for this set of 4 objects taken 2 at a time? Answer the question
(C) Using the two formulas for nPr
In Example 2 you probably found the multiplication principle to be the easiest

method to use. But for large values of n and r, you will find that the factorial formula
is more convenient. In fact, many calculators have functions that compute n! and nPr
directly.
EXAMPLE 3 Permutations Find the number of permutations of 13 objects taken 8 at a time.

Compute the answer using a calculator.
SOLUTION We use the factorial formula for nPr:
13! 13!
13P8 = = = 51,891,840
113 - 82! 5!
Using a tree diagram to solve this problem would involve a monumental effort.
Using the multiplication principle would mean multiplying 13 # 12 # 11 # 10 # 9 # 8 # 7 # 6
(8 factors), which is not too bad. However, a calculator can provide instant results
(see Fig. 1).
Matched Problem 3 Find the number of permutations of 30 objects taken 4 at

a time. Compute the answer using a calculator.
Figure 1
Combinations
Suppose that an art museum owns 8 paintings by a given artist and another art
museum wishes to borrow 3 of these paintings for a special show. In selecting 3 of
the 8 paintings for shipment, the order would not matter, and we would simply be se-
lecting a 3-element subset from the set of 8 paintings. That is, we would be selecting
what is called a combination of 8 objects taken 3 at a time.
DEFINITION Combination of n Objects Taken r at a Time

A combination of a set of n distinct objects taken r at a time without repetition is
an r-element subset of the set of n objects. The arrangement of the elements in the
subset does not matter.
How many ways can the 3 paintings be selected out of the 8 available? That
is, what is the number of combinations of 8 objects taken 3 at a time? To answer
this question, and to get a better insight into the general problem, we return to
Example 2.
In Example 2, we were given the set 5A, B, C6 and found the number of permu-
tations of 3 objects taken 2 at a time using a tree diagram. From this tree diagram, we
can determine the number of combinations of 3 objects taken 2 at a time (the number
of 2-element subsets from a 3-element set), and compare it with the number of per-
mutations (see Fig. 2).
FIRST SECOND
CHOICE CHOICE PERMUTATION
B AB
A
C AC
A BA
Start B
C BC
A CA
C
B CB
6 permutations of 3 objects
(A, B) (A, C) (B, C)
taken 2 at a time
(B, A) (C, A) (C, B) (Order does matter)
3 combinations of 3 objects
{A, B} {A, C} {B, C} taken 2 at a time
(Order does not matter)
Figure 2
There are fewer combinations than permutations, as we would expect. To each

subset (combination), there corresponds two ordered pairs (permutations). We denote
the number of combinations in Figure 2 by
3
3C2 or a b
2
Our final goal is to find a factorial formula for nCr, the number of combinations of
n objects taken r at a time. But first, we will develop a formula for 3C2, and then we
will generalize from this experience.
We know the number of permutations of 3 objects taken 2 at a time is given by
P
3 2 , and we have a formula for computing this number. Now, suppose we think of 3P2
in terms of two operations:
O1: Selecting a subset of 2 elements N1: 3C2 ways
O2: Arranging the subset in a given order N2: 2! ways

The combined operation, O1 followed by O2, produces a permutation of 3 objects

taken 2 at a time. Thus,
= 3C2 # 2!
3P2
3P2 or 3C2 =
2!
To find 3C2, the number of combinations of 3 objects taken 2 at a time, we substitute
3!
3P2 =
13 - 22!
and solve for 3C2:
3! 3#2#1
3C2
12 # 12112
= = = 3
2!13 ∙ 22!
This result agrees with the result obtained by using a tree diagram. Note that the
number of combinations of 3 objects taken 2 at a time is the same as the number of
permutations of 3 objects taken 2 at a time divided by the number of permutations of
the elements in a 2-element subset. Figure 2 also reveals this observation.
Reasoning the same way as in the example, the number of combinations of n
objects taken r at a time 10 … r … n2 is given by
nPr n!
nCr = Substitute nPr = .
r! 1n - r2!
n!
=
r!1n - r2!
THEOREM 3 Number of Combinations of n Objects Taken r at a Time

The number of combinations of n distinct objects taken r at a time without repeti-
tion is given by*
n 52
nCr = a b 52C5 = a b
r 5
P
n r 52P5
= =
r! 5!
n! 52!
= 0 … r … n =
r!1n - r2! 5!152 - 52!
n
* In place of the symbols nCr and a b, the symbols Cnr , Cn,r and C1n, r2 are often used.
r
Now we can answer the question posed earlier in the museum example. There are
8! 8! 8 # 7 # 6 # 5!
8C3 = = = # # # = 56
3!18 - 32! 3!5! 3 2 1 5!
ways that 3 paintings can be selected for shipment. That is, there are 56 combinations
of 8 objects taken 3 at a time.
EXAMPLE 4 Permutations and Combinations From a committee of 10 people,

(A) In how many ways can we choose a chairperson, a vice-chairperson, and a sec-
retary, assuming that one person cannot hold more than one position?
(B) In how many ways can we choose a subcommittee of 3 people?
SOLUTION Note how parts (A) and (B) differ. In part (A), order of choice makes a
difference in the selection of the officers. In part (B), the ordering does not matter
in choosing a 3-person subcommittee. In part (A), we are interested in the number

of permutations of 10 objects taken 3 at a time; and in part (B), we are interested
in the number of combinations of 10 objects taken 3 at a time. These quantities are
computed as follows (and since the numbers are not large, we do not need to use a
calculator):
10! 10! 10 # 9 # 8 # 7!
(A) 10 P3 = = = = 720 ways
110 - 32! 7! 7!
10! 10! 10 # 9 # 8 # 7!
3 # 2 # 1 # 7!
(B) 10C3 = = = = 120 ways
3!110 - 32! 3!7!
Matched Problem 4 From a committee of 12 people,

(A) In how many ways can we choose a chairperson, a vice-chairperson, a sec-
retary, and a treasurer, assuming that one person cannot hold more than one
position?
(B) In how many ways can we choose a subcommittee of 4 people?
If n and r are large numbers, a calculator is useful in evaluating expressions

involving factorials. Many calculators have a function that computes nCr directly
Figure 3 (see Fig. 3).
EXAMPLE 5 Combinations Find the number of combinations of 13 objects taken 8 at a time.

Compute the answer using a calculator.
13 13! 13!
SOLUTION 13C8 = a b = = = 1,287
8 8!113 - 82! 8!5!
Compare the result in Example 5 with that obtained in Example 3, and note that
13C8 is substantially smaller than 13P8 (see Fig. 3).
Matched Problem 5 Find the number of combinations of 30 objects taken 4 at

a time. Compute the answer using a calculator.
CONCEPTUAL INSIGHT
Permutations and combinations are similar in that both are selections in which
repetition is not allowed. But there is a crucial distinction between the two:
In a permutation, order is vital.
In a combination, order is irrelevant.
To determine whether a given selection is a permutation or combination, see if re-
arranging the order in which the elements are selected would produce a different
object. If so, the selection is a permutation; if not, the selection is a combination.

(A) List alphabetically by the first letter, all 3-letter license plate codes consisting of
3 different letters chosen from M, A, T, H. Discuss how this list relates to nPr.
(B) Reorganize the list from part (A) so that all codes without M come first, then
all codes without A, then all codes without T, and finally all codes without H.
Discuss how this list illustrates the formula nPr = r!nCr.
Applications
We now consider some applications of permutations and combinations. Several
applications in this section involve a standard 52-card deck of playing cards.
Standard 52-Card Deck of Playing Cards

A standard deck of 52 cards (see Fig. 4) has four 13-card suits: diamonds, hearts,
clubs, and spades. The diamonds and hearts are red, and the clubs and spades are
black. Each 13-card suit contains cards numbered from 2 to 10, a jack, a queen, a
king, and an ace. The number or letter on a card indicates its rank. So there are 13
ranks and 4 cards of each rank. The jack, queen, and king are called face cards. (The
ace is not a face card.) Depending on the game, the ace may be counted as the lowest
and/or the highest card in the suit. In traditional card games, a hand of cards is an
unordered subset of the deck.
Figure 4
EXAMPLE 6 Counting Techniques How many 5-card hands have 3 aces and 2 kings?
SOLUTION The solution involves both the multiplication principle and combina-
tions. Think of selecting the 5-card hand in terms of the following two operations:
O1: Choosing 3 aces out of 4 possible N1: 4C3
(order is not important)
O2: Choosing 2 kings out of 4 possible N2: 4C2
(order is not important)
Using the multiplication principle, we have
number of hands = 4C3 # 4C2
=
4! # 4!
3!14 - 32! 2!14 - 22!
= 4 # 6 = 24
Matched Problem 6 How many 5-card hands have 3 hearts and 2 spades?
EXAMPLE 7 Counting Techniques Serial numbers for a product are made using 2 letters
followed by 3 numbers. If the letters are taken from the first 8 letters of the alphabet
with no repeats and the numbers are taken from the 10 digits (0–9) with no repeats,
how many serial numbers are possible?
SOLUTION The solution involves both the multiplication principle and permuta-
tions. Think of selecting a serial number in terms of the following two operations:
O1: Choosing 2 letters out of 8 available N1: 8P2
(order is important)
O2: Choosing 3 numbers out of 10 available N2: 10P3
(order is important)
Using the multiplication principle, we have
number of serial numbers = 8P2 # 10P3
=
8! # 10!
18 - 22! 110 - 32!
= 56 # 720 = 40,320
Matched Problem 7 Repeat Example 7 under the same conditions, except

that the serial numbers will now have 3 letters followed by 2 digits (no repeats).
EXAMPLE 8 Counting Techniques A company has 7 senior and 5 junior officers. It wants to
form an ad hoc legislative committee. In how many ways can a 4-officer committee
be formed so that it is composed of
(A) Any 4 officers?
(B) 4 senior officers?
(C) 3 senior officers and 1 junior officer?
(D) 2 senior and 2 junior officers?
(E) At least 2 senior officers?
SOLUTION
(A) Since there are a total of 12 officers in the company, the number of different
4-member committees is
12! 12!
12C4 = = = 495
4!112 - 42! 4!8!
(B) If only senior officers can be on the committee, the number of different com-
mittees is
7! 7!
7C4 = = = 35
4!17 - 42! 4!3!
(C) The 3 senior officers can be selected in 7C3 ways, and the 1 junior officer can
be selected in 5C1 ways. Applying the multiplication principle, the number of
ways that 3 senior officers and 1 junior officer can be selected is
7C3
# 5C1 =
7! # 5!
=
7!5!
= 175
3!17 - 32! 1!15 - 12! 3!4!1!4!
(D) 7C2 # 5C2 = #

7! 5! 7!5!
= = 210
2!17 - 22! 2!15 - 22! 2!5!2!3!
(E) The committees with at least 2 senior officers can be divided into three disjoint
collections:
1. Committees with 4 senior officers and 0 junior officers
2. Committees with 3 senior officers and 1 junior officer
3. Committees with 2 senior officers and 2 junior officers
The number of committees of types 1, 2, and 3 is computed in parts (B), (C), and
(D), respectively. The total number of committees of all three types is the sum of
these quantities:
Type 1 Type 2 Type 3
# #
7C4 + 7C3 5C1 + 7C2 5C2 = 35 + 175 + 210 = 420
Matched Problem 8 Given the information in Example 8, answer the follow-

ing questions:
(A) How many 4-officer committees with 1 senior officer and 3 junior officers can
be formed?
(B) How many 4-officer committees with 4 junior officers can be formed?
(C) How many 4-officer committees with at least 2 junior officers can be formed?
EXAMPLE 9 Counting Techniques From a standard 52-card deck, how many 3-card hands
have all cards from the same suit?
SOLUTION There are 13 cards in each suit, so the number of 3-card hands having
all hearts, for example, is
13! 13!
13C3 = = = 286
3!113 - 32! 3!10!
Similarly, there are 286 3-card hands having all diamonds, 286 having all clubs, and
286 having all spades. So the total number of 3-card hands having all cards from
the same suit is
4 # 13C3 = 1,144
Matched Problem 9 From a standard 52-card deck, how many 5-card hands
have all cards from the same suit?
Exercises 6.4
Skills Warm-up Exercises A In Problems 7–26, evaluate the expression. If the answer is not an
integer, round to four decimal places.
W In Problems 1–6, evaluate the given expression without using a
calculator. (If necessary, review Section B.4). 7. 8! 8. 9!
12 # 11 # 10 12 # 10 # 8 9. 14 + 32! 10. 17 + 32!
3#2#1 6#4#2
1. 2.
11. 123 - 172! 12. 152 - 472!
10 # 9 # 8 # 7 # 6 8#7#6#5
5#4#3#2#1
3. 4. # # # 11! 20!
4 3 2 1 13. 14.
8! 18!
100 # 99 # 98 # c # 3 # 2 # 1
98 # 97 # 96 # c # 3 # 2 # 1
5. 8! 12!
15. 16.
12 # 11 # 10 # c # 3 # 2 # 1
4!18 - 42! 6!112 - 62!
9#8#7#6#5#4#3#2#1
6. 500! 1001!
17. 18.
498! 999!
19. 13C8 20. 15C10 44. From a standard 52-card deck, how many 5-card hands
consist entirely of queens?
21. 18P6 22. 10P7
45. From a standard 52-card deck, how many 7-card hands
12P7 365P25
23. 24. contain four kings?
7 25
12 365
39C5 26C4 consist of cards from the same suit?
25. 26.
52C5 52C4 47. From a standard 52-card deck, how many 4-card hands
In Problems 27–30, simplify each expression assuming that n is an contain a card from each suit?
integer and n Ú 2.
n! 1n + 12! consist of exactly one face card?
27. 28.
1n - 22! 3!1n - 22! 49. From a standard 52-card deck, how many 5-card hands
1n + 12! 1n + 22! contain 3 cards of one rank and 2 cards of a different rank?
29. 30.
2!1n - 12! 1n - 12! 50. From a standard 52-card deck, how many 5-card hands
contain 5 different ranks of cards?
In Problems 31–36, would you consider the selection to be a per-
mutation, a combination, or neither? Explain your reasoning. 51. A catering service offers 8 appetizers, 10 main courses,
and 7 desserts. A banquet committee selects 3 appetizers,
31. The university president named 3 new officers: a vice- 4 main courses, and 2 desserts. How many ways can this
president of finance, a vice-president of academic affairs, be done?
and a vice-president of student affairs.
52. Three committees from the House of Commons have 14, 16,
32. The university president selected 2 of her vice-presidents and 20 members, respectively. If each committee selects a
to attend the dedication ceremony of a new branch delegate and an alternate to represent the committee at a joint
campus. meeting, in how many ways can this be done?
33. A student checked out 4 novels from the library.
34. A student bought 4 books: 1 for his father, 1 for his mother, 1 In Problems 53 and 54, refer to the table in the graphing calcula-
for his younger sister, and 1 for his older brother. tor display below, which shows y1 = nPr and y2 = nCr for n = 6.
35. A father ordered an ice cream cone (chocolate, vanilla, or

strawberry) for each of his 4 children.
36. A book club meets monthly at the home of one of its 10
members. In December, the club selects a host for each meet-
ing of the next year.
37. In a horse race, how many different finishes among the
first 3 places are possible if 10 horses are running?
(Exclude ties.)
38. In a long-distance foot race, how many different finishes 53. Discuss and explain the symmetry of the numbers in the y2
among the first 5 places are possible if 50 people are run- column of the table.
ning? (Exclude ties.) 54. Explain how the table illustrates the formula
39. How many ways can a 3-person subcommittee be selected nPr = r!nCr
from a committee of 7 people? How many ways can a
president, vice-president, and secretary be chosen from a
In Problems 55–60, discuss the validity of each statement. If the
committee of 7 people?
statement is true, explain why. If not, give a counterexample.
40. Nine cards are numbered with the digits from 1 to 9. A 3-card
55. If n is a positive integer, then n! 6 1n + 12!
hand is dealt, 1 card at a time. How many hands are possible
in which 56. If n is a positive integer greater than 3, then n! 7 2n.
(A) Order is taken into consideration? 57. If n and r are positive integers and 1 6 r 6 n, then
(B) Order is not taken into consideration? nPr 6 nPr + 1.
B 41. From a standard 52-card deck, how many 6-card hands 58. If n and r are positive integers and 1 6 r 6 n, then
consist entirely of red cards? nCr 6 nCr + 1.
42. From a standard 52-card deck, how many 6-card hands 59. If n and r are positive integers and 1 6 r 6 n, then
consist entirely of clubs? nCr = nCn - r.
43. From a standard 52-card deck, how many 5-card hands 60. If n and r are positive integers and 1 6 r 6 n, then
consist entirely of face cards? nPr = nPn - r.
C 61. Eight distinct points are selected on the circumference of a 70. Note from the table in the graphing calculator dis-
circle. play that the largest value of nCr when n = 21 is
(A) How many line segments can be drawn by joining the 21C10 = 21C11 = 352,716. Use a similar table to find the
points in all possible ways? largest value of nCr when n = 19.
(B) How many triangles can be drawn using these 8 points as

vertices?
(C) How many quadrilaterals can be drawn using these 8 points
as vertices?
62. Five distinct points are selected on the circumference of a
circle.
(A) How many line segments can be drawn by joining the
points in all possible ways?
(B) How many triangles can be drawn using these 5 points as
vertices?
(C) How many quadrilaterals can be drawn using these
5 points as vertices?
Applications
63. In how many ways can 4 people sit in a row of 6 chairs? 71. Quality control. An office supply store receives a shipment
64. In how many ways can 3 people sit in a row of 8 chairs? of 24 high-speed printers, including 5 that are defective.
Three of these printers are selected for a store display.
65. A basketball team has 5 distinct positions. Out of 8 players,
how many starting teams are possible if (A) How many selections can be made?
(A) The distinct positions are taken into consideration? (B) How many of these selections will contain no defective
printers?
(B) The distinct positions are not taken into consideration?
72. Quality control. An electronics store receives a shipment
(C) The distinct positions are not taken into consideration,
of 30 graphing calculators, including 6 that are defective.
but either Mike or Ken (but not both) must start?
Four of these calculators are selected for a local high
66. How many 4-person committees are possible from a group of school.
9 people if
(A) There are no restrictions? (A) How many selections can be made?
(B) Both Jim and Mary must be on the committee? (B) How many of these selections will contain no defective
calculators?
(C) Either Jim or Mary (but not both) must be on the com-
mittee? 73. Business closings. A clothing retail company with 14 stores
in France, 15 in Italy, and 11 in Germany is planning to close
67. Let U be the set of all 2-card hands, let K be the set of all
8 of these stores.
2-card hands that contain exactly 1 king, and let H be the
set of all 2-card hands that contain exactly 1 heart. Find (A) In how many ways can this be done?
n1K ¨ H′2, n1K ¨ H2, n1K′ ¨ H2, and n1K′ ¨ H′2.
(B) The company decides to close 3 stores in France, 4 in
68. Let U be the set of all 2-card hands, let K be the set of all Italy, and 1 in Germany. In how many ways can this be
2-card hands that contain exactly 1 king, and let Q be the done?
set of all 2-card hands that contain exactly 1 queen. Find
74. Employee layoffs. A real estate company with 14 employ-
n1K ¨ Q′2, n1K ¨ Q2, n1K′ ¨ Q2, and n1K′ ¨ Q′2.
ees in its central office, 8 in its north office, and 6 in its south
69. Note from the table in the graphing calculator display below office is planning to lay off 12 employees.
that the largest value of nCr when n = 20 is 20C10 = 184,756.
(A) How many ways can this be done?
Use a similar table to find the largest value of nCr when
n = 24. (B) The company decides to lay off 5 employees from the
central office, 4 from the north office, and 3 from the
south office. In how many ways can this be done?
75. Personnel selection. Suppose that 6 female and 5 male
applicants have been successfully screened for 5 positions.
In how many ways can the following compositions be
selected?
(A) 3 females and 2 males
(B) 4 females and 1 male
(C) 5 females Answers to Matched Problems

(D) 5 people regardless of sex 1. (A) 720 (B) 10 (C) 720 (D) 20 (E) 1,140
2. (A)
(E) At least 4 females B AB
A C AC
76. Committee selection. A 4-person grievance committee is se- D AD
A BA
lected out of 2 departments A and B, with 15 and 20 people, B C BC
D BD
respectively. In how many ways can the following commit- Start A CA
tees be selected? C B CB
D CD
A DA
(A) 3 from A and 1 from B D B DB
C DC
(B) 2 from A and 2 from B 12 permutations of 4 objects taken 2 at a time
(C) All from A
(B) O1: Fill first position N1: 4 ways
(D) 4 people regardless of department O2: Fill second position N2: 3 ways
(E) At least 3 from department A 4 # 3 = 12
(C) 4P2 = 4 # 3 = 12; 4P2 =

77. Medicine. There are 8 standard classifications of blood 4!
= 12
type. An examination for prospective laboratory techni- 14 - 22!
cians consists of having each candidate determine the 30!
type for 3 blood samples. How many different examina- 3. 30P4 = = 657,720
130 - 42!
tions can be given if no 2 of the samples provided for the
candidate have the same type? If 2 or more samples have 12!
4. (A) 12P4 = = 11,880 ways
the same type? 112 - 42!
78. Medical research. Because of limited funds, 5 research 12!
(B) 12C4 = = 495 ways
centers are chosen out of 8 suitable ones for a study on heart 4!112 - 42!
disease. How many choices are possible? 30!
5. 30C4 = = 27,405
79. Politics. A nominating convention will select a president and 4!130 - 42!
6. 13C3 # 13C2 = 22,308
vice-president from among 4 candidates. Campaign buttons,
listing a president and a vice-president, will be designed for
each possible outcome before the convention. How many dif- 7. 8P3 # 10P2 = 30,240
ferent kinds of buttons should be designed? 8. (A) 7C1 # 5C3 = 70 (B) 5C4 = 5
80. Politics. In how many different ways can 6 candidates for an
# #
(C) 7C2 5C2 + 7C1 5C3 + 5C4 = 285
office be listed on a ballot? 9. 4 # 13C5 = 5,148

6.1 Logic EXAMPLES
• A proposition is a statement (not a question or command) that is either true or false.
• If p and q are propositions, then the compound propositions, Ex. 1, p. 332
¬p, p ¡ q, p ¿ q, and p S q
can be formed using the negation symbol ¬ and the connectives ¡, ¿ , and S . These
propositions are called not p, p or q, p and q, and if p then q, respectively (or negation,
disjunction, conjunction, and conditional, respectively). Each of these compound
propositions is specified by a truth table (see pages 330 and 331).
• Given any conditional proposition p S q, the proposition q S p is called the converse of p S q, Ex. 2, p. 332
and the proposition ¬q S ¬p, is called the contrapositive of p S q.
• A truth table for a compound proposition specifies whether it is true or false for any assignment Ex. 3, p. 333
of truth values to its variables. A proposition is a tautology if each entry in its column of the truth Ex. 4, p. 333
table is T, a contradiction if each entry is F, and a contingency if at least one entry is T and at least Ex. 5, p. 334
one entry is F.
• Consider the rows of the truth tables for the compound propositions P and Q. If whenever P is Ex. 6, p. 335
true, Q is also true, we say that P logically implies Q and write P 1 Q. We call P 1 Q a logical Ex. 7, p. 336
implication. If the compound propositions P and Q have identical truth tables, we say that P and Q
are logically equivalent and write P K Q. We call P K Q a logical equivalence.
• Several logical equivalences are given in Table 2 on page 336. The last of these implies that any
conditional proposition is logically equivalent to its contrapositive.
6.2 Sets
• A set is a collection of objects specified in such a way that we can tell whether any given object is
or is not in the collection.
• Each object in a set is called a member, or element, of the set. If a is an element of the set A, we
write a ∈ A.
• A set without any elements is called the empty, or null, set, denoted by ∅.
• A set can be described by listing its elements, or by giving a rule that determines the elements of Ex. 1, p. 339
the set. If P1x2 is a statement about x, then 5x ∙ P1x2 6 denotes the set of all x such that P1x2 is
true.
• A set is finite if its elements can be counted and there is an end; a set such as the positive integers,
in which there is no end in counting its elements, is infinite.
• We write A ⊂ B, and say that A is a subset of B, if each element of A is an element of B. We write Ex. 2, p. 340
A = B, and say that sets A and B are equal, if they have exactly the same elements. The empty set Ex. 3, p. 340
∅ is a subset of every set.
• If A and B are sets, then Ex. 4, p. 342
A ∪ B = 5x ∙ x H A or x H B6
is called the union of A and B, and
A ¨ B = 5x ∙ x H A and x H B6
is called the intersection of A and B.
• Venn diagrams are useful in visualizing set relationships. Ex. 5, p. 342
• If A ¨ B = ∅, the sets A and B are said to be disjoint.
• The set of all elements under consideration in a given discussion is called the universal set U. The
set A= = 5x H U∙ x ∙ A6 is called the complement of A (relative to U).
• The number of elements in set A is denoted by n1A2. So if A and B are sets, then the numbers
that are often shown in a Venn diagram, as in Figure 4 on page 341, are n1A ¨ B = 2, n1A ¨ B2,
n1B ¨ A= 2, and n1A= ¨ B = 2 (see Fig. 9 on page 342).
6.3 Basic Counting Principles
• If A and B are sets, then the number of elements in the union of A and B is given by the addition Ex. 1, p. 347
principle for counting (Theorem 1, page 347). Ex. 2, p. 348
• If the elements of a set are determined by a sequence of operations, tree diagrams can be used to list Ex. 3, p. 349
all combined outcomes. To count the number of combined outcomes without using a tree diagram, Ex. 4, p. 351
use the multiplication principle for counting (Theorem 2, page 350). Ex. 5, p. 351
6.4 Permutations and Combinations

• The product of the first n natural numbers, denoted n!, is called n factorial: Ex. 1, p. 355
n! = n1n - 12 1n - 22 # g# 2 # 1
0! = 1
n! = n # 1n - 12!
• A permutation of a set of distinct objects is an arrangement of the objects in a specific order without Ex. 2, p. 357
repetition. The number of permutations of a set of n distinct objects is given by nPn = n!. A permuta- Ex. 3, p. 358
tion of a set of n distinct objects taken r at a time without repetition is an arrangement of r of the n
objects in a specific order. The number of permutations of n distinct objects taken r at a time without
repetition is given by
n!
n Pr = 0 … r … n
1n - r2!
• A combination of a set of n distinct objects taken r at a time without repetition is an r-element sub- Ex. 4, p. 360
set of the set of n objects. The arrangement of the elements in the subset is irrelevant. The number Ex. 5, p. 361
of combinations of n distinct objects taken r at a time without repetition is given by Ex. 6, p. 362
Ex. 7, p. 363
n n Pr n!
n Cr = a b = = 0 … r … n Ex. 8, p. 363
r r! r!1n - r2!
Ex. 9, p. 364
Review Exercises
Work through all the problems in this chapter review and check In Problems 17–19, write the resulting set using the listing
your answers in the back of the book. Answers to all problems are method.
there along with section numbers in italics to indicate where each 17. 51, 2, 3, 46 ∪ 52, 3, 4, 56
type of problem is discussed. Where weaknesses show up, review
18. 51, 2, 3, 46 ¨ 52, 3, 4, 56
appropriate sections in the text.
19. 51, 2, 3, 46 ¨ 55, 66
In Problems 1–6, express each proposition in an English sentence
and determine whether it is true or false, where p and q are the 20. A single die is rolled, and a coin is flipped. How many
propositions combined outcomes are possible? Solve:
p: “23 6 32” q: “34 6 43”
1. ¬ p 2. p ¡ q
21. Use the Venn diagram to find the number of elements in
3. p ¿ q 4. p S q each of the following sets:
5. The converse of 6. q S p (A) A (B) B (C) U (D) A′
qSp
U
In Problems 7–10, indicate true (T) or false (F). A B
7. 5a, b, c6 = 5c, b, a6
8. 5s, d6 ⊂ 5a, b, c, d, s6 30 35 40
9. k H 5hike6
45
10. ∅ ⊂ 51, 2, 3, 46
Figure for Problems 21 and 22
In Problems 11–14, describe each proposition as a negation, dis- 22. Use the Venn diagram to find the number of elements in
junction, conjunction, or conditional, and determine whether the each of the following sets:
proposition is true or false.
(A) A ¨ B (B) A ∪ B
11. If 4 is composite, then 9 is even.
(C) 1A ¨ B2 ′ (D) 1A ∪ B2 ′
12. 63 is prime or 8 is even.
Evaluate the expressions in Problems 23–28.
13. 53 is prime and 57 is prime.
15!
23. 110 - 62! 24.
14. 51 is not prime. 10!
In Problems 15–16, state the converse and the contrapositive of 15!
25. 26. 8C5
the given proposition. 10!5!
15. If the square matrix A has a row of zeros, then the square 27. 8P5 28. #
13C4 13C1
matrix A has no inverse.
29. How many seating arrangements are possible with
16. If the square matrix A is an identity matrix, then the square 6 people and 6 chairs in a row? Solve using the multiplica-
matrix A has an inverse. tion principle.
30. Solve Problem 29 using permutations or combinations, (B) How many of the selections in part (A) consist entirely
whichever is applicable. of women?
In Problems 31–36, construct a truth table for the proposition and (C) How many ways can the department select a team of
determine whether the proposition is a contingency, tautology, or 3 programmers to work on a particular project?
contradiction. 53. A group of 150 people includes 52 who play chess, 93 who
play checkers, and 28 who play both chess and checkers.
31. 1p S q2 ¿ 1q S p2 32. p ¡ 1q S p2
How many people in the group play neither game?
33. 1p ¡ ¬p2 S 1q ¿ ¬q2 34. ¬q ¿ 1p S q2
Problems 54 and 55 refer to the following Venn diagram.
35. ¬p S 1p S q2 36. ¬1p ¡ ¬q2
U
In Problems 37–40, determine whether the given set is finite or in- A B
finite. Consider the set Z of integers to be the universal set, and let
x y z
M = 5n ∊ Z ∙ n 6 106 6
K = 5n ∊ Z ∙ n 7 103 6 w
E = 5n ∊ Z ∙ n is even6
54. Which of the numbers x, y, z, or w must equal 0 if A ⊂ B?
37. E ∪ K 38. M ¨ K 55. Which of the numbers x, y, z, or w must equal 0 if
39. K′ 40. E ¨ M A ¨ B = U?
In Problems 41–42, determine whether or not the given sets are In Problems 56–58, discuss the validity of each statement. If the
disjoint. For the definitions of M, K, and E, refer to the instruc- statement is true, explain why. If not, give a counterexample.
tions for Problems 37–40. 56. If n and r are positive integers and 1 6 r 6 n, then
41. M′ and K′ nCr 6 nPr .
42. M and E′ 57. If n and r are positive integers and 1 6 r 6 n, then nPr 6 n!.
43. Draw a Venn diagram for sets A, B, and C and shade the 58. If n and r are positive integers and 1 6 r 6 n, then nCr 6 n!.
region described by A′ ¨ 1B ∪ C2.
In Problems 59–64, construct a truth table to verify the implica-
44. A man has 5 children. Each of those children has 3 children, tion or equivalence.
who in turn each have 2 children. Discuss the number of
descendants that the man has. 59. p ¿ q 1 p 60. q 1 p S q
45. How many 3-letter code words are possible using the first 8 61. ¬p S 1q ¿ ¬q2 K p 62. p ¡ q K ¬p S q
letters of the alphabet if no letter can be repeated? If letters
63. p ¿ 1p S q2 1 q 64. ¬1p ¿ ¬q2 K p S q
can be repeated? If adjacent letters cannot be alike?
65. How many different 5-child families are possible where the
46. Solve the following problems using nPr or nCr:
gender of the children in the order of their births is taken into
(A) How many 3-digit opening combinations are possible on consideration [that is, birth sequences such as (B, G, G, B, B)
a combination lock with 6 digits if the digits cannot be and (G, B, G, B, B) produce different families]? How many
repeated? families are possible if the order pattern is not taken into ac-
(B) Five tennis players have made the finals. If each of the count?
5 players is to play every other player exactly once, how
66. Can a selection of r objects from a set of n distinct objects,
many games must be scheduled?
where n is a positive integer, be a combination and a permu-
47. Use graphical techniques on a graphing calculator to find the tation simultaneously? Explain.
largest value of nCr when n = 25.
48. If 3 operations O1, O2, O3 are performed in order, with pos-
sible number of outcomes N1, N2, N3, respectively, determine
the number of branches in the corresponding tree diagram.
Applications
In Problems 49–51, write the resulting set using the listing
method. 67. Transportation. A distribution center A wishes to send its
products to five different retail stores: B, C, D, E, and F. How
49. 5x ∙ x3 - x = 06
many different route plans can be constructed so that a single
50. 5x ∙x is a positive integer and x! 6 1006 truck, starting from A, will deliver to each store exactly once
51. 5x ∙ x is a positive integer that is a perfect square and x 6 506 and then return to the center?
52. A software development department consists of 6 women and 68. Market research. A survey of 1,000 people indicates that
4 men. 340 have invested in stocks, 480 have invested in bonds, and
(A) How many ways can the department select a chief 210 have invested in stocks and bonds.
programmer, a backup programmer, and a programming (A) How many people in the survey have invested in stocks
librarian? or bonds?
(B) How many have invested in neither stocks nor bonds? advantage that might accrue to candidates whose names
appear near the top of the ballot, it is proposed that equal
(C) How many have invested in bonds and not stocks?
numbers of ballots be printed for each possible order in
69. Medical research. In a study of twins, a sample of 6 pairs of which the candidates’ names can be listed.
identical twins will be selected for medical tests from a group
(A) In how many ways can the candidates’ names be listed?
of 40 pairs of identical twins. In how many ways can this be
done? (B) Explain why the proposal is not feasible, and discuss
possible alternatives.
70. Elections. In an unusual recall election, there are 67
candidates to replace the governor of a state. To negate the
7 Probability
7.1 Sample Spaces, Events,

and Probability
Introduction
Like other branches of mathematics, probability evolved out of practical
7.2 Union, Intersection, and
Complement of Events; considerations. Girolamo Cardano (1501–1576), a gambler and physician,
Odds produced some of the best mathematics of his time, including a systematic
analysis of gambling problems. In 1654, another gambler, Chevalier de
7.3 Conditional Probability,
Méré, approached the well-known French philosopher and mathematician
Intersection, and
Independence Blaise Pascal (1623–1662) regarding certain dice problems. Pascal became
interested in these problems, studied them, and discussed them with Pierre de
7.4 Bayes’ Formula Fermat (1601–1665), another French mathematician. So out of the gaming
7.5 Random Variable, rooms of western Europe, the study of probability was born.
Probability Distribution, Despite this lowly birth, probability has matured into a highly respected and
and Expected Value immensely useful branch of mathematics. It is used in practically every field.
In particular, probability plays a critical role in the management of risk, from
insuring a sculpture against theft (see Problem 49 in Section 7.5) to diversifying
an investment portfolio to designing a health care system.
372
SECTION 7.1 Sample Spaces, Events, and Probability 373
7.1 Sample Spaces, Events, and Probability

■■ Experiments This section provides a brief and relatively informal introduction to probability. It
■■ Sample Spaces and Events
assumes a familiarity with the basics of set theory as presented in Chapter 6, includ-
ing the union, intersection, and complement of sets, as well as various techniques
■■ Probability of an Event for counting the number of elements in a set. Probability studies involve many
■■ Equally Likely Assumption subtle ideas, and care must be taken at the beginning to understand the fundamental
concepts.
Experiments
Some experiments do not yield the same results each time that they are performed, no
matter how carefully they are repeated under the same conditions. These experiments
are called random experiments. Familiar examples of random experiments are flip-
ping coins, rolling dice, observing the frequency of defective items from an assembly
line, or observing the frequency of deaths in a certain age group.
Probability theory is a branch of mathematics that has been developed to deal
with outcomes of random experiments, both real and conceptual. In the work that fol-
lows, we simply use the word experiment to mean a random experiment.
Sample Spaces and Events

Associated with outcomes of experiments are sample spaces and events. Consider the
experiment, “A wheel with 18 numbers on the perimeter (Fig. 1) spins and comes to
rest so that a pointer points within a numbered sector.”
What outcomes might we observe? When the wheel stops, we might be interested
Figure 1 in which number is next to the pointer, or whether that number is an odd number, or
whether that number is divisible by 5, or whether that number is prime, or whether
the pointer is in a red or green sector, and so on. The list of possible outcomes ap-
pears endless. In general, there is no unique method of analyzing all possible out-
comes of an experiment. Therefore, before conducting an experiment, it is important
to decide just what outcomes are of interest.
Suppose we limit our interest to the set of numbers on the wheel and to various
subsets of these numbers, such as the set of prime numbers or the set of odd numbers
on the wheel. Having decided what to observe, we make a list of outcomes of the
experiment, called simple outcomes or simple events, such that in each trial of the
experiment (each spin of the wheel), one and only one of the outcomes on the list will
occur. For our stated interests, we choose each number on the wheel as a simple event
and form the set
S = 51, 2, 3, c, 17, 186
The set of simple events S for the experiment is called a sample space for the
experiment.
Now consider the outcome, “When the wheel comes to rest, the number next to
the pointer is divisible by 4.” This outcome is not a simple outcome (or simple event)
since it is not associated with one and only one element in the sample space S. The
outcome will occur whenever any one of the simple events 4, 8, 12, or 16 occurs, that
is, whenever an element in the subset
E = 54, 8, 12, 166
occurs. Subset E is called a compound event (and the outcome, a compound
outcome).
374 CHAPTER 7 Probability
DEFINITION Sample Spaces and Events

If we formulate a set S of outcomes (events) of an experiment in such a way that
in each trial of the experiment one and only one of the outcomes (events) in the
set will occur, we call the set S a sample space for the experiment. Each element
in S is called a simple outcome, or simple event.
An event E is defined to be any subset of S (including the empty set ∅ and
the sample space S). Event E is a simple event if it contains only one element and
a compound event if it contains more than one element. We say that an event E
occurs if any of the simple events in E occurs.
We use the terms event and outcome of an experiment interchangeably.

Technically, an event is the mathematical counterpart of an outcome of an experi-
ment, but we will not insist on strict adherence to this distinction in our development
of probability.
Real World Mathematical Model

Experiment (real or conceptual) Sample space (set S)
Outcome (simple or compound) Event (subset of S; simple or compound)
EXAMPLE 1 Simple and Compound Events Relative to the number wheel experiment (Fig. 1)
and the sample space
S = 51, 2, 3, c, 17, 186
what is the event E (subset of the sample space S) that corresponds to each of the
following outcomes? Indicate whether the event is a simple event or a compound
event.
(A) The outcome is a prime number. (B) The outcome is the square of 4.
SOLUTION
(A) The outcome is a prime number if any of the simple events 2, 3, 5, 7, 11, 13,
or 17 occurs.* To say “A prime number occurs” is the same as saying that the
experiment has an outcome in the set
E = 52, 3, 5, 7, 11, 13, 176
Since event E has more than one element, it is a compound event.
(B) The outcome is the square of 4 if 16 occurs. To say “The square of 4 occurs” is
the same as saying that the experiment has an outcome in the set
E = 5166
Since E has only one element, it is a simple event.

(A) The outcome is a number divisible by 12.
(B) The outcome is an even number greater than 15.
*Technically, we should write 526, 536, 556, 576, 5116, 5136, and 5176 for the simple events since
there is a logical distinction between an element of a set and a subset consisting of only that element. But
we will keep this in mind and drop the braces for simple events to simplify the notation.
EXAMPLE 2 Sample Spaces A nickel and a dime are tossed. How do we identify a sample
space for this experiment? There are a number of possibilities, depending on our
interest. We will consider three.
(A) If we are interested in whether each coin falls heads (H) or tails (T), then, using
a tree diagram, we can easily determine an appropriate sample space for the
experiment:
NICKLE DIME COMBINED
OUTCOMES OUTCOMES OUTCOMES
H HH
H
T HT
Start
H TH
T
T TT
S1 = 5HH, HT, TH, TT6

and there are 4 simple events in the sample space.
(B) If we are interested only in the number of heads that appear on a single toss of
the two coins, we can let
S2 = 50, 1, 26
and there are 3 simple events in the sample space.
(C) If we are interested in whether the coins match (M) or do not match (D), we
can let
S3 = 5M, D6
and there are only 2 simple events in the sample space.
CONCEPTUAL INSIGHT
There is no single correct sample space for a given experiment. When specify-
ing a sample space for an experiment, we include as much detail as necessary to
answer all questions of interest regarding the outcomes of the experiment. If in
doubt, we choose a sample space that contains more elements rather than fewer.
In Example 2, which sample space would be appropriate for all three inter-
ests? Sample space S1 contains more information than either S2 or S3. If we know
which outcome has occurred in S1, then we know which outcome has occurred in
S2 and S3. However, the reverse is not true. (Note that the simple events in S2 and
S3 are simple or compound events in S1.) In this sense, we say that S1 is a more
fundamental sample space than either S2 or S3. Thus, we would choose S1 as an
appropriate sample space for all three expressed interests.
Matched Problem 2 An experiment consists of recording the boy–girl com-

position of a two-child family. What would be an appropriate sample space
(A) If we are interested in the genders of the children in the order of their births?
Draw a tree diagram.
(B) If we are interested only in the number of girls in a family?
(C) If we are interested only in whether the genders are alike (A) or different (D)?
(D) For all three interests expressed in parts (A) to (C)?
EXAMPLE 3 Sample Spaces and Events Consider an experiment of rolling two dice. Figure 2
shows a convenient sample space that will enable us to answer many questions
about interesting events. Let S be the set of all ordered pairs in the figure. The sim-
ple event (3, 2) is distinguished from the simple event (2, 3). The former indicates
that a 3 turned up on the first die and a 2 on the second, while the latter indicates
that a 2 turned up on the first die and a 3 on the second.
Second Die
(1, 1) (1, 2) (1, 3) (1, 4) (1, 5) (1, 6)
(2, 1) (2, 2) (2, 3) (2, 4) (2, 5) (2, 6)
(3, 1) (3, 2) (3, 3) (3, 4) (3, 5) (3, 6)

First Die
(4, 1) (4, 2) (4, 3) (4, 4) (4, 5) (4, 6)
(5, 1) (5, 2) (5, 3) (5, 4) (5, 5) (5, 6)
(6, 1) (6, 2) (6, 3) (6, 4) (6, 5) (6, 6)

Figure 2
What is the event (subset of the sample space S) that corresponds to each of the
following outcomes?
(A) A sum of 7 turns up. (B) A sum of 11 turns up.
(C) A sum less than 4 turns up. (D) A sum of 12 turns up.
SOLUTION
(A) By “A sum of 7 turns up,” we mean that the sum of all dots on both turned-up
faces is 7. This outcome corresponds to the event
5 16, 12, 15, 22, 14, 32, 13, 42, 12, 52, 11, 626
(B) “A sum of 11 turns up” corresponds to the event
5 16, 52, 15, 626
(C) “A sum less than 4 turns up” corresponds to the event
5 11, 12, 12, 12, 11, 226
(D) “A sum of 12 turns up” corresponds to the event
5 16, 626
Matched Problem 3 Refer to the sample space shown in Figure 2. What is the
event that corresponds to each of the following outcomes?
(A) A sum of 5 turns up.
(B) A sum that is a prime number greater than 7 turns up.
As indicated earlier, we often use the terms event and outcome of an experiment
interchangeably. In Example 3, we might say “the event, ‘A sum of 11 turns up’” in
place of “the outcome, ‘A sum of 11 turns up,’” or even write
E = a sum of 11 turns up = 5 16, 52, 15, 626
Probability of an Event
The next step in developing our mathematical model for probability studies is the
introduction of a probability function. This is a function that assigns to an arbitrary
event associated with a sample space a real number between 0 and 1, inclusive. We
start by discussing ways in which probabilities are assigned to simple events in the
sample space S.
DEFINITION Probabilities for Simple Events

Given a sample space
S = 5e1, e2, c, en 6
with n simple events, to each simple event ei we assign a real number, denoted by
P1ei 2, called the probability of the event ei. These numbers can be assigned in an
arbitrary manner as long as the following two conditions are satisfied:
Condition 1. The probability of a simple event is a number between 0 and 1, inclu-
sive. That is,
0 … P1ei 2 … 1
Condition 2. The sum of the probabilities of all simple events in the sample space
is 1. That is,
P1e1 2 + P1e2 2 + g + P1en 2 = 1
Any probability assignment that satisfies Conditions 1 and 2 is said to be an accept-
able probability assignment.
Our mathematical theory does not explain how acceptable probabilities are as-
signed to simple events. These assignments are generally based on the expected or
actual percentage of times that a simple event occurs when an experiment is repeated
a large number of times. Assignments based on this principle are called reasonable.
Let an experiment be the flipping of a single coin, and let us choose a sample
space S to be
S = 5H, T6
If a coin appears to be fair, we are inclined to assign probabilities to the simple events
in S as follows:
1 1
P1H2 = and P1T2 =
2 2
These assignments are based on reasoning that, since there are 2 ways a coin can
land, in the long run, a head will turn up half the time and a tail will turn up half the
time. These probability assignments are acceptable since both conditions for accept-
able probability assignments stated in the preceding box are satisfied:
1. 0 … P1H2 … 1, 0 … P1T2 … 1
1 1
2. P1H2 + P1T2 = + = 1
2 2
If we were to flip a coin 1,000 times, we would expect a head to turn up
approximately, but not exactly, 500 times. The random number feature on a graph-
ing calculator can be used to simulate 1,000 flips of a coin. Figure 3 shows the
results of 3 such simulations: 497 heads the first time, 495 heads the second, and
Figure 3 504 heads the third.
If, however, we get only 376 heads in 1,000 flips of a coin, we might suspect that
the coin is not fair. Then we might assign the simple events in the sample space S the
following probabilities, based on our experimental results:
P1H2 = .376 and P1T2 = .624
This is also an acceptable assignment. However, the probability assignment
P1H2 = 1 and P1T2 = 0
although acceptable, is not reasonable (unless the coin has 2 heads). And the
assignment
P1H2 = .6 and P1T2 = .8
is not acceptable, since .6 + .8 = 1.4, which violates Condition 2 in the box on
page 377.*
It is important to keep in mind that out of the infinitely many possible acceptable
probability assignments to simple events in a sample space, we are generally inclined
to choose one assignment over another based on reasoning or experimental results.
Given an acceptable probability assignment for simple events in a sample space
S, how do we define the probability of an arbitrary event E associated with S?
DEFINITION Probability of an Event E

Given an acceptable probability assignment for the simple events in a sample space
S, we define the probability of an arbitrary event E, denoted by P1E2, as follows:
(A) If E is the empty set, then P1E2 = 0.
(B) If E is a simple event, then P1E2 has already been assigned.
(C) If E is a compound event, then P1E2 is the sum of the probabilities of all the
simple events in E.
(D) If E is the sample space S, then P1E2 = P1S2 = 1 [this is a special case of
part (C)].
EXAMPLE 4 Probabilities of Events Let us return to Example 2, the tossing of a nickel and a
dime, and the sample space
S = 5HH, HT, TH, TT6
Since there are 4 simple outcomes and the coins are assumed to be fair, it would
appear that each outcome would occur 25% of the time, in the long run. Let us as-
sign the same probability of 14 to each simple event in S:
Simple Event
ei HH HT TH TT
1 1 1 1
P1ei 2
4 4 4 4
This is an acceptable assignment according to Conditions 1 and 2, and it is a rea-

sonable assignment for ideal (perfectly balanced) coins or coins close to ideal.
(A) What is the probability of getting 1 head (and 1 tail)?
(B) What is the probability of getting at least 1 head?
(C) What is the probability of getting at least 1 head or at least 1 tail?
(D) What is the probability of getting 3 heads?
*In probability studies, the 0 to the left of the decimal is usually omitted. So we write .6 and .8 instead of
0.6 and 0.8.
SOLUTION
(A) E1 = getting 1 head = 5HT, TH6
Since E1 is a compound event, we use part (C) in the box and find P1E1 2 by
adding the probabilities of the simple events in E1:
1 1 1
P1E1 2 = P1HT2 + P1TH2 = + =
4 4 2
(B) E2 = getting at least 1 head = 5HH, HT, TH6
1 1 1 3
P1E2 2 = P1HH2 + P1HT2 + P1TH2 = + + =
4 4 4 4
(C) E3 = 5HH, HT, TH, TT6 = S
1 1 1 1
P1E3 2 = P1S2 = 1 4 + 4 + 4 + 4 = 1
(D) E4 = getting 3 heads = ∅ Empty set
P1∅ 2 = 0
PROCEDURE Steps for Finding the Probability of an Event E
Step 1 Set up an appropriate sample space S for the experiment.

Step 2 Assign acceptable probabilities to the simple events in S.
Step 3 To obtain the probability of an arbitrary event E, add the probabilities of the
simple events in E.
The function P defined in Steps 2 and 3 is a probability function whose

domain is all possible events (subsets) in the sample space S and whose range is a
set of real numbers between 0 and 1, inclusive.
Matched Problem 4 Suppose in Example 4 that after flipping the nickel and
dime 1,000 times, we find that HH turns up 273 times, HT turns up 206 times, TH
turns up 312 times, and TT turns up 209 times. On the basis of this evidence, we
assign probabilities to the simple events in S as follows:
Simple Event
ei HH HT TH TT
P1ei 2 .273 .206 .312 .209
This is an acceptable and reasonable probability assignment for the simple events in
S. What are the probabilities of the following events?
(A) E1 = getting at least 1 tail
(B) E2 = getting 2 tails
(C) E3 = getting at least 1 head or at least 1 tail
Example 4 and Matched Problem 4 illustrate two important ways in which

acceptable and reasonable probability assignments are made for simple events in a
sample space S. Each approach has its advantage in certain situations:
1. Theoretical Approach. We use assumptions and a deductive reasoning process
to assign probabilities to simple events. No experiments are actually conducted.
This is what we did in Example 4.
2. Empirical Approach. We assign probabilities to simple events based on the results
of actual experiments. This is what we did in Matched Problem 4.
Empirical probability concepts are stated more precisely as follows: If we con-

duct an experiment n times and event E occurs with frequency f1E2, then the ratio
f1E2 >n is called the relative frequency of the occurrence of event E in n trials. We
define the empirical probability of E, denoted by P1E2, by the number (if it exists)
that the relative frequency f1E2 >n approaches as n gets larger and larger. Therefore,
frequency of occurrence of E f1E2

P1E2 ≈ =
total number of trials n
For any particular n, the relative frequency f1E2 >n is also called the approximate
empirical probability of event E.
For most of this section, we emphasize the theoretical approach. In the next sec-
tion, we return to the empirical approach.
Equally Likely Assumption

In tossing a nickel and a dime (Example 4), we assigned the same probability, 14, to
each simple event in the sample space S = 5HH, HT, TH, TT6. By assigning the
same probability to each simple event in S, we are actually making the assumption
that each simple event is as likely to occur as any other. We refer to this as an equally
likely assumption. If, in a sample space
S = 5e1, e2, c, en 6
with n elements, we assume that each simple event ei is as likely to occur as any
other, then we assign the probability 1>n to each. That is,
1
P1ei 2 =
n
Under an equally likely assumption, we can develop a very useful formula for
finding probabilities of arbitrary events associated with a sample space S. Consider
the following example:
If a single die is rolled and we assume that each face is as likely to come up as
any other, then for the sample space
S = 51, 2, 3, 4, 5, 66
1
we assign a probability of to each simple event since there are 6 simple events. The
6
probability of
E = rolling a prime number = 52, 3, 56
is Number of elements in E
T
1 1 1 3 1
P1E2 = P122 + P132 + P152 = + + = =
6 6 6 6 2
c
Number of elements in S
Under the assumption that each simple event is as likely to occur as any other, the
computation of the probability of the occurrence of any event E in a sample space S is
the number of elements in E divided by the number of elements in S.
THEOREM 1 Probability of an Arbitrary Event under an Equally Likely

Assumption
If we assume that each simple event in sample space S is as likely to occur as any
other, then the probability of an arbitrary event E in S is given by
number of elements in E n1E2

P1E2 = =
number of elements in S n1S2
EXAMPLE 5 Probabilities and Equally Likely Assumptions Let us again consider rolling two
dice, and assume that each simple event in the sample space shown in Figure 2
(page 376) is as likely as any other. Find the probabilities of the following events:
(A) E1 = a sum of 7 turns up
(B) E2 = a sum of 11 turns up
(C) E3 = a sum less than 4 turns up
(D) E4 = a sum of 12 turns up
SOLUTION Referring to Figure 2 (page 376) and the results found in Example 3,
we find
n1E1 2 6 1
(A) P1E1 2 = = = E1 = 516, 12, 15, 22, 14, 32, 13, 42, 12, 52, 11, 626
n1S2 36 6
n1E2 2 2 1
(B) P1E2 2 = = = E2 = 516, 52, 15, 62 6
n1S2 36 18
n1E3 2 3 1
(C) P1E3 2 = = = E3 = 511, 12, 12, 12, 11, 22 6
n1S2 36 12
n1E4 2 1
(D) P1E4 2 = = E4 = 516, 62 6
n1S2 36
Matched Problem 5 Under the conditions in Example 5, find the probabilities

of the following events (each event refers to the sum of the dots facing up on both
dice):
(B) E6 = a sum that is a prime number greater than 7 turns up
EXAMPLE 6 Simulation and Empirical Probabilities Use output from the random number
feature of a graphing calculator to simulate 100 rolls of two dice. Determine the
empirical probabilities of the following events, and compare with the theoretical
probabilities:
SOLUTION A graphing calculator can be used to select a random integer from 1 to 6.
Each of the six integers in the given range is equally likely to be selected. Therefore,
by selecting a random integer from 1 to 6 and adding it to a second random integer
from 1 to 6, we simulate rolling two dice and recording the sum (see the first com-
mand in Figure 4A). The second command in Figure 4A simulates 100 rolls of two
dice; the sums are stored in list L 1. From the statistical plot of L 1 in Figure 4B we
obtain the empirical probabilities.*
15
(A) The empirical probability of E1 is 100 = .15; the theoretical probability of E1
6
(see Example 5A) is 36 = .167.
6
(B) The empirical probability of E2 is 100 = .06; the theoretical probability of E2
2
(see Example 5B) is 36 = .056.
*If you simulate this experiment on your graphing calculator, you should not expect to get the same empirical
probabilities.
20
0
22 14
Figure 4 (A) (B)
Matched Problem 6 Use the graphing calculator output in Figure 4B to

determine the empirical probabilities of the following events, and compare with
the theoretical probabilities:
(A) E3 = a sum less than 4 turns up

A shipment box contains 12 graphing calculators, out of which 2 are defective. A
calculator is drawn at random from the box and then, without replacement, a second
calculator is drawn. Discuss whether the equally likely assumption would be appro-
priate for the sample space S = 5GG, GD, DG, DD6, where G is a good calculator
and D is a defective one.
We now turn to some examples that make use of the counting techniques devel-
oped in Chapter 6.
EXAMPLE 7 Probability and Equally Likely Assumption In drawing 5 cards from a 52-card
deck without replacement, what is the probability of getting 5 spades?
SOLUTION Let the sample space S be the set of all 5-card hands from a 52-card
deck. Since the order in a hand does not matter, n1S2 = 52C5. Let event E be
the set of all 5-card hands from 13 spades. Again, the order does not matter and
n1E2 = 13C5. Assuming that each 5-card hand is as likely as any other,
n1E2 13C5 1,287

P1E2 = = = ≈ .0005
n1S2 52C5 2,598,960
(Some calculators display the answer as 4.951980792E–4. This means the same
thing as the scientific notation 4.951980792 * 10 -4, so the answer, rounded to
4 decimal places, is .0005).
Matched Problem 7 In drawing 7 cards from a 52-card deck without replace-

ment, what is the probability of getting 7 hearts?
EXAMPLE 8 Probability and Equally Likely Assumption The board of regents of a university
is made up of 12 men and 16 women. If a committee of 6 is chosen at random, what
is the probability that it will contain 3 men and 3 women?
SOLUTION Let S be the set of all 6-person committees out of 28 people. Then
n1S2 = 28C6
Let E be the set of all 6-person committees with 3 men and 3 women. To find n1E2,
we use the multiplication principle and the following two operations:
O1: Select 3 men out of the 12 available N1: 12C3
O2: Select 3 women out of the 16 available N2: 16C3
Therefore,
n1E2 = N1 # N2 = 12C3
# 16C3
and
n1E2 #
12C3 16C3
P1E2 = = ≈ .327
n1S2 28C6
Matched Problem 8 What is the probability that the committee in Example 8

will have 4 men and 2 women?
There are many counting problems for which it is not possible to produce a
simple formula that will yield the number of possible cases. In situations of this type,
we often revert back to tree diagrams and counting branches.
Exercises 7.1
Skills Warm-up Exercises Refer to the description of a standard deck of 52 cards and Figure 4
on page 362. An experiment consists of drawing 1 card from a
W In Problems 1–6, without using a calculator, determine which standard 52-card deck. In Problems 15–24, what is the probability
event, E or F, is more likely to occur. (If necessary, review of drawing
Section B.1.)
15. A club 16. A black card
5 4
1. P1E2 = ; P1F2 = 17. A heart or diamond 18. A numbered card
6 5
19. The jack of clubs 20. An ace
2 1
2. P1E2 = ; P1F2 = 21. An ace or red card 22. A red queen
7 3
3 23. A black diamond 24. A six or club
3. P1E2 = ; P1F2 = .4
8 25. In a family with 2 children, excluding multiple births, what is
7 the probability of having 2 children of the opposite gender?
4. P1E2 = .9; P1F2 =
8 Assume that a girl is as likely as a boy at each birth.
1 26. In a family with 2 children, excluding multiple births, what
5. P1E2 = .15; P1F2 =
6 is the probability of having 2 girls? Assume that a girl is as
6 8 likely as a boy at each birth.
6. P1E2 = ; P1F2 =
7 9 27. A store carries four brands of DVD players: J, G, P, and S.
From past records, the manager found that the relative
A A circular spinner is divided into 15 sectors of equal area: 6 red frequency of brand choice among customers varied. Which
sectors, 5 blue, 3 yellow, and 1 green. In Problems 7–14, consider of the following probability assignments for a particular cus-
the experiment of spinning the spinner once. Find the probability tomer choosing a particular brand of DVD player would have
that the spinner lands on: to be rejected? Why?
7. Blue 8. Red (A) P1J2 = .15, P1G2 = - .35, P1P2 = .50, P1S2 = .70
9. Yellow or green 10. Red or blue (B) P1J2 = .32, P1G2 = .28, P1P2 = .24, P1S2 = .30
11. Orange 12. Yellow, red, or green (C) P1J2 = .26, P1G2 = .14, P1P2 = .30, P1S2 = .30
13. Blue, red, yellow, or green 14. Purple 28. Using the probability assignments in Problem 27C, what is the
probability that a random customer will not choose brand S?
29. Using the probability assignments in Problem 27C, what is 45. Sum is less than 5. 46. Sum is greater than 8.
the probability that a random customer will choose brand G
47. Sum is not 7 or 11. 48. Sum is not 2, 4, or 6.
or brand J?
49. Sum is 1. 50. Sum is 13.
30. Using the probability assignments in Problem 27C, what is
the probability that a random customer will not choose brand 51. Sum is divisible by 3. 52. Sum is divisible by 4.
G or brand J?
53. Sum is 7 or 11 (a “natural”). 54. Sum is 2, 3, or 12 (“craps”).
B 31. In a family with 3 children, excluding multiple births, what
55. Sum is divisible by 2 or 3. 56. Sum is divisible by 2 and 3.
is the probability of having 2 boys and 1 girl, in that order?
Assume that a boy is as likely as a girl at each birth. An experiment consists of tossing three fair (not weighted) coins,
32. In a family with 3 children, excluding multiple births, what except that one of the three coins has a head on both sides. Compute
is the probability of having 2 boys and 1 girl, in any order? the probability of obtaining the indicated results in Problems 57–62.
Assume that a boy is as likely as a girl at each birth. 57. 1 head 58. 2 heads
33. A keypad at the entrance of a building has 10 buttons labeled 59. 3 heads 60. 0 heads
0 through 9. What is the probability of a person correctly
guessing a 4-digit entry code? 61. More than 1 head 62. More than 1 tail
34. A keypad at the entrance of a building has 10 buttons labeled 0 In Problems 63–68, a sample space S is described. Would it be
through 9. What is the probability of a person correctly guess- reasonable to make the equally likely assumption? Explain.
ing a 4-digit entry code if they know that no digits repeat?
63. A single card is drawn from a standard deck. We are interest-
Refer to the description of a standard deck of 52 cards and Figure 4 ed in whether or not the card drawn is a heart, so an appropri-
on page 362. An experiment consists of dealing 5 cards from a ate sample space is S = 5H, N6.
standard 52-card deck. In Problems 35–38, what is the probability 64. A single fair coin is tossed. We are interested in whether the
of being dealt coin falls heads or tails, so an appropriate sample space is
35. 5 black cards? 36. 5 hearts? S = 5H, T6.
37. 5 face cards? 38. 5 nonface cards? 65. A single fair die is rolled. We are interested in whether or not
the number rolled is even or odd, so an appropriate sample
39. Twenty thousand students are enrolled at a state university. space is S = 5E, O6.
A student is selected at random, and his or her birthday
(month and day, not year) is recorded. Describe an appropri- 66. A nickel and dime are tossed. We are interested in the num-
ate sample space for this experiment and assign acceptable ber of heads that appear, so an appropriate sample space is
probabilities to the simple events. What are your assumptions S = 50, 1, 26.
in making this assignment? 67. A wheel of fortune has seven sectors of equal area col-
40. In a local authority mayor election, polls indicate that two ored red, orange, yellow, green, blue, indigo, and violet.
of the three candidates are running neck-and-neck, while the We are interested in the color that the pointer indicates
third candidate is receiving only one-third of the support of when the wheel stops, so an appropriate sample space is
either of the others. Registered voters are chosen at random S = 5R, O, Y, G, B, I, V6.
and asked which of the three will get their vote. Describe an 68. A wheel of fortune has eight sectors of equal area colored
appropriate sample space for this random survey experiment blue, green, blue, red, blue, green, blue, and red. We are inter-
and assign acceptable probabilities to the simple events. ested in the color that the pointer indicates when the wheel
41. Suppose that 5 thank-you notes are written and 5 envelopes stops, so an appropriate sample space is S = 5B, G, R6.
are addressed. Accidentally, the notes are randomly inserted 69. (A) Is it possible to get 19 heads in 20 flips of a fair coin?
into the envelopes and mailed without checking the address- Explain.
es. What is the probability that all the notes will be inserted
into the correct envelopes? (B) If you flipped a coin 40 times and got 37 heads, would
you suspect that the coin was unfair? Why or why not?
42. Suppose that 6 people check their coats in a cloakroom. If all If you suspect an unfair coin, what empirical probabili-
claim checks are lost and the 6 coats are randomly returned, what ties would you assign to the simple events of the sample
is the probability that all the people will get their own coats back? space?
An experiment consists of rolling two fair dice and adding the 70. (A) Is it possible to get 7 double 6’s in 10 rolls of a pair of
dots on the two sides facing up. Using the sample space shown in fair dice? Explain.
Figure 2 (page 376) and, assuming each simple event is as likely (B) If you rolled a pair of dice 36 times and got 11 double
as any other, find the probability of the sum of the dots indicated 6’s, would you suspect that the dice were unfair? Why
in Problems 43–56. or why not? If you suspect loaded dice, what empirical
43. Sum is 6. 44. Sum is 8. probability would you assign to the event of rolling a
double 6?
C An experiment consists of rolling two fair (not weighted) 4-sided (B) What is the probability that a 6 is rolled under the
dice and adding the dots on the two sides facing up. Each die is equally likely assumption?
numbered 1–4. Compute the probability of obtaining the indicated
(C) Use a graphing calculator to simulate 100 rolls of a fair
sums in Problems 71–78.
die and determine the empirical probabilities of the six
71. 2 72. 3 73. 4 outcomes.
74. 5 75. 6 76. 7 88. Use a graphing calculator to simulate 200 tosses of a nickel
and dime, representing the outcomes HH, HT, TH, and TT by
77. An odd sum 78. An even sum 1, 2, 3, and 4, respectively.
In Problems 79–86, find the probability of being dealt the given (A) Find the empirical probabilities of the four outcomes.
hand from a standard 52-card deck. Refer to the description of a (B) What is the probability of each outcome under the
standard 52-card deck on page 362. equally likely assumption?
79. A 5-card hand that consists entirely of red cards 89. (A) Explain how a graphing calculator can be used to simu-
late 500 tosses of a coin.
80. A 5-card hand that consists entirely of face cards
(B) Carry out the simulation and find the empirical prob-
81. A 6-card hand that contains exactly three face cards
abilities of the two outcomes.
82. A 6-card hand that contains exactly two clubs
(C) What is the probability of each outcome under the
83. A 4-card hand that contains no aces equally likely assumption?
84. A 4-card hand that contains no face cards 90. From a box containing 12 balls numbered 1 through 12, one
ball is drawn at random.
85. A 7-card hand that contains exactly 2 diamonds and exactly
(A) Explain how a graphing calculator can be used to simu-
2 spades
late 400 repetitions of this experiment.
86. A 7-card hand that contains exactly 1 king and exactly
(B) Carry out the simulation and find the empirical probabil-
2 jacks
ity of drawing the 8 ball.
In Problems 87–90, several experiments are simulated using the (C) What is the probability of drawing the 8 ball under the
random number feature on a graphing calculator. For example, the equally likely assumption?
roll of a fair die can be simulated by selecting a random integer from
1 to 6, and 50 rolls of a fair die by selecting 50 random integers from
1 to 6 (see Fig. A for Problem 87 and your user’s manual).
87. From the statistical plot of the outcomes of rolling a fair die Applications
50 times (see Fig. B), we see, for example, that the number 4
was rolled exactly 5 times. 91. Consumer testing. Twelve popular brands of fruit juices are
used in a blind taste study for consumer recognition.
(A) If 4 distinct brands are chosen at random from the 12
and if a consumer is not allowed to repeat any answers,
what is the probability that all 4 brands could be identi-
fied by just guessing?
(B) If repeats are allowed in the 4 brands chosen at random
from the 12 and if a consumer is allowed to repeat an-
swers, what is the probability that all 4 brands are identi-
fied correctly by just guessing?
(A)
92. Consumer testing. Six popular brands of cola are to be used
in a blind taste study for consumer recognition.
14
(A) If 3 distinct brands are chosen at random from the 6 and
if a consumer is not allowed to repeat any answers, what
is the probability that all 3 brands could be identified by
just guessing?
(B) If repeats are allowed in the 3 brands chosen at random
0
0 8 from the 6 and if a consumer is allowed to repeat an-
swers, what is the probability that all 3 brands are identi-
(B) fied correctly by just guessing?
(A) What is the empirical probability that the number 6 was
93. Personnel selection. Suppose that 6 female and 5 male
rolled?
applicants have been successfully screened for 5 positions.
If the 5 positions are filled at random from the 11 finalists, (B) What is the probability that a majority of committee
what is the probability of selecting members are Conservatives?
(A) 3 females and 2 males? 98. Politics. A temporary joint committee of the UK Parliament
has 22 members: 10 members of the House of Lords (half
(B) 4 females and 1 male?
Conservative, half Labor) and 12 members of the House of
(C) 5 females? Commons (half Conservative, half Labor). A 6-person sub-
committee is chosen at random from all the members of the
(D) At least 4 females?
Joint Committee.
94. Committee selection. A 4-person grievance committee is
(A) What is the probability that the sub-committee contains
to include employees in 2 departments, A and B, with 15
equal number of members from both houses of Parliament?
and 20 employees, respectively. If the 4 people are selected
at random from the 35 employees, what is the probability of (B) What is the probability that the sub-committee contains
selecting equal number of Conservative members and Labor
members?
(A) 3 from A and 1 from B?
(B) 2 from A and 2 from B? Answers to Matched Problems
(C) All 4 from A? 1. (A) E = 5126; simple event
(B) E = 516, 186; compound event
(D) At least 3 from A?
2. (A) S1 = 5BB, BG, GB, GG6;
95. Medicine. A laboratory technician is to be tested on identify-
ing blood types from 8 standard classifications. GENDER OF GENDER OF COMBINED
FIRST CHILD SECOND CHILD OUTCOMES
(A) If 3 distinct samples are chosen at random from the 8
types and if the technician is not allowed to repeat any B BB
B
answers, what is the probability that all 3 could be cor- G BG
Start
rectly identified by just guessing? B GB
G
G GG
(B) If repeats are allowed in the 3 blood types chosen at ran-
dom from the 8 and if the technician is allowed to repeat (B) S2 = 50, 1, 26 (C) S3 = 5A, D6 (D) S1
answers, what is the probability that all 3 are identified 3. (A) 514, 12, 13, 22, 12, 32, 11, 42 6
correctly by just guessing?
(B) 516, 52, 15, 62 6
96. Medical research. Because of limited funds, 5 research 4. (A) .727 (B) .209 (C) 1
centers are to be chosen out of 8 suitable ones for a study on 1 1
heart disease. If the selection is made at random, what is the 5. (A) P1E5 2 = (B) P1E6 2 =
9 18
probability that 5 particular research centers will be chosen? 9 3
6. (A) = .09 1empirical2; = .083 1theoretical2
97. Politics. A town council in the United Kingdom has 11 100 36
members: 6 Conservative and 5 Labor. A 3-person committee 1 1
(B) = .01 1empirical2; = .028 1theoretical2
is selected at random. 100 36
(A) What is the probability that all committee members are 7. ≈ 1.3 * 10-5
13C7 > 52C7
Conservatives? 8. C #
12 4 16 2 28C6 ≈ .158
C >
7.2 Union, Intersection, and Complement of Events; Odds

■■ Union and Intersection Recall from Section 7.1 that given a sample space
■■ Complement of an Event S = 5e1, e2, c , en 6
■■ Odds any function P defined on S such that
■■ Applications to Empirical Probability 0 … P1ei 2 … 1 i = 1, 2, c , n
and
P1e1 2 + P1e2 2 + g + P1en 2 = 1
is called a probability function. In addition, any subset of S is called an event E, and
the probability of E is the sum of the probabilities of the simple events in E.
SECTION 7.2 Union, Intersection, and Complement of Events; Odds 387
Union and Intersection

Since events are subsets of a sample space, the union and intersection of events are sim-
ply the union and intersection of sets as defined in the following box. In this section, we
concentrate on the union of events and consider only simple cases of intersection. More
complicated cases of intersection will be investigated in the next section.
DEFINITION Union and Intersection of Events*

If A and B are two events in a sample space S, then the union of A and B, denoted by
A ∪ B, and the intersection of A and B, denoted by A ¨ B, are defined as follows:
A ∪ B = 5e ∊ S ∙ e ∊ A or e ∊ B6 A ¨ B = 5e ∊ S ∙ e ∊ A and e ∊ B6
S S
A B A B
A < B is shaded A > B is shaded
Furthermore, we define
The event A or B to be A ∪ B
The event A and B to be A ¨ B
*See Section 6.2 for a discussion of set notation.
EXAMPLE 1 Probability Involving Union and Intersection Consider the sample space of
equally likely events for the rolling of a single fair die:
S = 51, 2, 3, 4, 5, 66
(A) What is the probability of rolling a number that is odd and exactly divisible by 3?
(B) What is the probability of rolling a number that is odd or exactly divisible by 3?
SOLUTION
(A) Let A be the event of rolling an odd number, B the event of rolling a number
divisible by 3, and F the event of rolling a number that is odd and divisible
by 3. Then (Fig. 1A)
A = 51, 3, 56 B = 53, 66 F = A ¨ B = 536
S The probability of rolling a number that is odd and exactly divisible by 3 is
F5A>B
A B
n1A ¨ B2 1
1 2 3 P1F2 = P1A ¨ B2 = =
n1S2 6
(B) Let A and B be the same events as in part (A), and let E be the event of rolling a
4 5 6
number that is odd or divisible by 3. Then (Fig. 1B)
A = 51, 3, 56 B = 53, 66 E = A ∪ B = 51, 3, 5, 66
(A)
The probability of rolling a number that is odd or exactly divisible by 3 is
S
n1A ∪ B2 4 2
1 2 3
P1E2 = P1A ∪ B2 = = =
n1S2 6 3
E5A<B
Matched Problem 1 Use the sample space in Example 1 to answer the
4 5 6
following:
(A) What is the probability of rolling an odd number and a prime number?
(B)
(B) What is the probability of rolling an odd number or a prime number?
Figure 1
Suppose that A and B are events in a sample space S. How is the probability of
A ∪ B related to the individual probabilities of A and of B? Think of the probability
of an element of S as being its weight. To find the total weight of the elements of
A ∪ B, we weigh all of the elements of A, then weigh all of the elements of B, and
subtract from the sum the weight of all of the elements that were weighed twice. This
gives the formula for P1A ∪ B2 in Theorem 1.
THEOREM 1 Probability of the Union of Two Events

For any events A and B,
P1A ∪ B2 = P1A2 + P1B2 - P1A ¨ B2 (1)
S Events A and B are mutually exclusive (disjoint) if their intersection is the

A B
empty set (Fig. 2). In that case, equation (1) simplifies, because the probability of the
empty set is 0. So, if A ¨ B = ∅, then P1A ∪ B2 = P1A2 + P1B2.
CONCEPTUAL INSIGHT
Figure 2 Mutually exclusive: Note the similarity between equation (1) in Theorem 1 and the formula obtained
A¨B = ∅ in Section 6.3 for counting the number of elements in A ∪ B:
n1A ∪ B2 = n1A2 + n1B2 - n1A ¨ B2
To find the number of elements in A ∪ B, we count all of the elements of A, then
count all of the elements of B, and subtract from the sum the number of elements
that were counted twice.
EXAMPLE 2 Probability Involving Union and Intersection Suppose that two fair dice are rolled.
(A) What is the probability that a sum of 7 or 11 turns up?
(B) What is the probability that both dice turn up the same or that a sum less than
5 turns up?
SOLUTION
(A) If A is the event that a sum of 7 turns up and B is the event that a sum of 11
turns up, then (Fig. 3) the event that a sum of 7 or 11 turns up is A ∪ B, where
A = 5 11, 62, 12, 52, 13, 42, 14, 32, 15, 22, 16, 126
B = 5 15, 62, 16, 526
Second Die
(1, 1) (1, 2) (1, 3) (1, 4) (1, 5) (1, 6)

A
(2, 1) (2, 2) (2, 3) (2, 4) (2, 5) (2, 6)
(3, 1) (3, 2) (3, 3) (3, 4) (3, 5) (3, 6)

First Die
(4, 1) (4, 2) (4, 3) (4, 4) (4, 5) (4, 6)
(5, 1) (5, 2) (5, 3) (5, 4) (5, 5) (5, 6)

B
(6, 1) (6, 2) (6, 3) (6, 4) (6, 5) (6, 6)
Figure 3
Since events A and B are mutually exclusive, we can use the simplified version
of equation (1) to calculate P1A ∪ B2:
P1A ∪ B2 = P1A2 + P1B2 In this equally likely sample space,
6 2 n1A2 = 6, n1B2 = 2, and n1S2 = 36.
= +
36 36
8 2
= =
36 9
(B) If A is the event that both dice turn up the same and B is the event that the sum
is less than 5, then (Fig. 4) the event that both dice turn up the same or the sum
is less than 5 is A ∪ B, where
A = 5 11, 12, 12, 22, 13, 32, 14, 42, 15, 52, 16, 626
B = 5 11, 12, 11, 22, 11, 32, 12, 12, 12, 22, 13, 126
Since A ¨ B = 5 11, 12, 12, 226, A and B are not mutually exclusive, and we
use equation (1) to calculate P1A ∪ B2:
P1A ∪ B2 = P1A2 + P1B2 - P1A ¨ B2
6 6 2
= + -
36 36 36
10 5
= =
36 18
Second Die
(1, 1) (1, 2) (1, 3) (1, 4) (1, 5) (1, 6)

A>B B
(2, 1) (2, 2) (2, 3) (2, 4) (2, 5) (2, 6)
(3, 1) (3, 2) (3, 3) (3, 4) (3, 5) (3, 6)

First Die
(4, 1) (4, 2) (4, 3) (4, 4) (4, 5) (4, 6)
(5, 1) (5, 2) (5, 3) (5, 4) (5, 5) (5, 6)

A
(6, 1) (6, 2) (6, 3) (6, 4) (6, 5) (6, 6)
Figure 4
Matched Problem 2 Use the sample space in Example 2 to answer the

following:
(A) What is the probability that a sum of 2 or 3 turns up?
(B) What is the probability that both dice turn up the same or that a sum greater
than 8 turns up?
You no doubt noticed in Example 2 that we actually did not have to use equation
(1). We could have proceeded as in Example 1 and simply counted sample points in
A ∪ B. The following example illustrates the use of equation (1) in a situation where
visual representation of sample points is not practical.
EXAMPLE 3 Probability Involving Union and Intersection What is the probability that a num-
ber selected at random from the first 500 positive integers is (exactly) divisible by
3 or 4?
SOLUTION Let A be the event that a drawn integer is divisible by 3 and B the event
that a drawn integer is divisible by 4. Note that events A and B are not mutually
exclusive because multiples of 12 are divisible by both 3 and 4. Since each of the
positive integers from 1 to 500 is as likely to be drawn as any other, we can use
n1A2, n1B2, and n1A ¨ B2 to determine P1A ∪ B2, where
500
n1A2 = the largest integer less than or equal to = 166
3
500
n1B2 = the largest integer less than or equal to = 125
4
500
n1A ¨ B2 = the largest integer less than or equal to = 41
12
Now we can compute P1A ∪ B2:
P1A ∪ B2 = P1A2 + P1B2 - P1A ¨ B2
n1A2 n1B2 n1A ¨ B2
= + -
n1S2 n1S2 n1S2
166 125 41 250
= + - = = .5
500 500 500 500
Matched Problem 3 What is the probability that a number selected at random

from the first 140 positive integers is (exactly) divisible by 4 or 6?
Complement of an Event
Suppose that we divide a finite sample space
S = 5e1, c, en 6
into two subsets E and E′ such that
S
E ¨ E′ = ∅
E
E9 that is, E and E′ are mutually exclusive and
E ∪ E′ = S
Then E′ is called the complement of E relative to S. Note that E′ contains all the ele-
Figure 5 ments of S that are not in E (Fig. 5). Furthermore,
P1S2 = P1E ∪ E′2
= P1E2 + P1E′2 = 1
Therefore,
P1E2 = 1 - P1E′2 P1E′2 = 1 - P1E2 (2)
If the probability of rain is .67, then the probability of no rain is 1 - .67 = .33;
if the probability of striking oil is .01, then the probability of not striking oil is .99.
If the probability of having at least 1 boy in a 2-child family is .75, what is the prob-
ability of having 2 girls? [Answer:.25.]

(A) Suppose that E and F are complementary events. Are E and F necessarily mutu-
ally exclusive? Explain why or why not.
(B) Suppose that E and F are mutually exclusive events. Are E and F necessarily
complementary? Explain why or why not.
In looking for P1E2, there are situations in which it is easier to find P1E′2 first,
and then use equations (2) to find P1E2. The next two examples illustrate two such
situations.
EXAMPLE 4 Quality Control A shipment of 45 precision parts, including 9 that are defective,
is sent to an assembly plant. The quality control division selects 10 at random for
testing and rejects the entire shipment if 1 or more in the sample are found to be
defective. What is the probability that the shipment will be rejected?
SOLUTION If E is the event that 1 or more parts in a random sample of 10 are defec-
tive, then E′, the complement of E, is the event that no parts in a random sample of
10 are defective. It is easier to compute P1E′2 than to compute P1E2 directly. Once
P1E′2 is found, we will use P1E2 = 1 - P1E′2 to find P1E2.
The sample space S for this experiment is the set of all subsets of 10 elements
from the set of 45 parts shipped. Thus, since there are 45 - 9 = 36 nondefective
parts,
n1E′2 36C10
P1E′2 = = ≈ .08
n1S2 45C10
and
P1E2 = 1 - P1E′2 ≈ 1 - .08 = .92
Matched Problem 4 A shipment of 40 precision parts, including 8 that are

defective, is sent to an assembly plant. The quality control division selects 10 at
random for testing and rejects the entire shipment if 1 or more in the sample are
found to be defective. What is the probability that the shipment will be rejected?
EXAMPLE 5 Birthday Problem In a group of n people, what is the probability that at least
2 people have the same birthday (same month and day, excluding February 29)?
Make a guess for a class of 40 people, and check your guess with the conclusion of
this example.
SOLUTION If we form a list of the birthdays of all the people in the group, we have
a simple event in the sample space
S = set of all lists of n birthdays
For any person in the group, we will assume that any birthday is as likely as any other,
so that the simple events in S are equally likely. How many simple events are in the set
S? Since any person could have any one of 365 birthdays (excluding February 29), the
multiplication principle implies that the number of simple events in S is
1st 2nd 3rd nth
person person person person
n1S2 = 365 # 365 # 365 # g # 365
= 365n
Now, let E be the event that at least 2 people in the group have the same birthday.
Then E′ is the event that no 2 people have the same birthday. The multiplication
principle can be used to determine the number of simple events in E′:
1st 2nd 3rd nth
person person person person
n1E′2 = 365 # 364 # 363 # g # 1366 - n2
3365 # 364 # 363 # g # 1366 - n241365 - n2! Multiply numerator
= and denominator by
1365 - n2! 1365 - n2!.
365!
=
1365 - n2!
Since we have assumed that S is an equally likely sample space,

1
365!
n1E′2 1365 - n2! 365!
P1E′2 = = n = n
n1S2 365 365 1365 - n2!
Therefore,
0 39
0
P1E2 = 1 - P1E′2 (3)
Figure 6 365!
= 1 - n
365 1365 - n2!
Table 1 Birthday Problem Equation ( 3) is valid for any n satisfying 1 … n … 365. [What is P1E2 if n 7 365?]
For example, in a group of 5 people,
Number Probability That
of People 2 or More Have 365!
in Group Same Birthday P1E2 = 1 -
n P1E2
13652 5 360!
5 .027 365 # 364 # 363 # 362 # 361 # 360!
365 # 365 # 365 # 365 # 365 # 360!
= 1 -
10 .117
15 .253 = .027
20 .411
23 .507 It is interesting to note that as the size of the group increases, P1E2 increas-
30 .706 es more rapidly than you might expect. Figure 6* shows the graph of P1E2 for
40 .891 1 … n … 39. Table 1 gives the value of P1E2 for selected values of n. If n = 5,
50 .970 Table 1 gives P1E2 = .027, as calculated above. Notice that for a group of only
60 .994 23 people, the probability that 2 or more have the same birthday is greater than 12.
70 .999
Matched Problem 5 Use equation (3) to evaluate P1E2 for n = 4.

Determine the smallest number n such that in a group of n people, the probability that
2 or more have a birthday in the same month is greater than .5. Discuss the assump-
tions underlying your computation.
Odds
When the probability of an event E is known, it is often customary (especially in
gaming situations) to speak of odds for or against E rather than the probability of E.
For example, if you roll a fair die once, then the odds for rolling a 2 are 1 to 5 (also
written 1 : 5), and the odds against rolling a 2 are 5 to 1 (or 5 : 1). This is consistent
with the following instructions for converting probabilities to odds.
*See Problem 71 in Exercises 7.2 for a discussion of the form of equation (3) used to produce the graph
in Figure 6.
DEFINITION From Probabilities to Odds

If P1E2 is the probability of the event E, then
P1E2 P1E2
(A) Odds for E = = P1E2 ∙ 1
1 - P1E2 P1E′2
P1E′2
(B) Odds against E = P1E2 ∙ 0
P1E2
P1E2 a
The ratio , giving odds for E, is usually expressed as an equivalent ratio of
P1E′2 b
whole numbers (by multiplying numerator and denominator by the same number),
and written “a to b” or “a : b.” In this case, the odds against E are written “b to a” or
“b : a.”
Odds have a natural interpretation in terms of fair games. Let’s return to the ex-
periment of rolling a fair die once. Recall that the odds for rolling a 2 are 1 to 5. Turn
the experiment into a fair game as follows: If you bet $1 on rolling a 2, then the house
pays you $5 (and returns your $1 bet) if you roll a 2; otherwise, you lose the $1 bet.
More generally, consider any experiment and an associated event E. If the odds
for E are a to b, then the experiment can be turned into a fair game as follows: If you
bet $a on event E, then the house pays you $b (and returns your $a bet) if E occurs;
otherwise, you lose the $a bet.
EXAMPLE 6 Probability and Odds

(A) What are the odds for rolling a sum of 7 in a single roll of two fair dice?
(B) If you bet $1 on rolling a sum of 7, what should the house pay (plus returning
your $1 bet) if you roll a sum of 7 in order for the game to be fair?
SOLUTION
6
(A) Let E denote the event of rolling a sum of 7. Then P1E2 = 36 = 16. So the
odds for E are
1
P1E2 6 1
= 5
= Also written as “1 to 5” or “1 : 5.”
P1E′2 6
5
(B) The odds for rolling a sum of 7 are 1 to 5. The house should pay $5 (and return
your $1 bet) if you roll a sum of 7 for the game to be fair.
Matched Problem 6
(A) What are the odds for rolling a sum of 8 in a single roll of two fair dice?
(B) If you bet $5 that a sum of 8 will turn up, what should the house pay (plus re-
turning your $5 bet) if a sum of 8 does turn up in order for the game to be fair?
Now we will go in the other direction: If we are given the odds for an event, what
is the probability of the event? (The verification of the following formula is left to
Problem 75 in Exercises 7.2.)
If the odds for event E are a , b, then the probability of E is
a
P1E2 ∙
a ∙ b
EXAMPLE 7 Odds and Probability If in repeated rolls of two fair dice, the odds for roll-
ing a 5 before rolling a 7 are 2 to 3, then the probability of rolling a 5 before
rolling a 7 is
a 2 2
P1E2 = = =
a + b 2 + 3 5
Matched Problem 7 If in repeated rolls of two fair dice, the odds against roll-
ing a 6 before rolling a 7 are 6 to 5, then what is the probability of rolling a 6 before
rolling a 7?
Applications to Empirical Probability

In the following discussions, the term empirical probability will mean the
probability of an event determined by a sample that is used to approximate the
probability of the corresponding event in the total population. How does the ap-
proximate empirical probability of an event determined from a sample relate to
the actual probability of an event relative to the total population? In mathematical
statistics an important theorem called the law of large numbers (or the law of
averages) is proved. Informally, it states that the approximate empirical prob-
ability can be made as close to the actual probability as we please by making the
sample sufficiently large.
EXAMPLE 8 Market Research From a survey of 1,000 people in Springfield, it was found that
500 people had tried a certain brand of diet cola, 600 had tried a certain brand of
regular cola, and 200 had tried both types of cola. If a person from Springfield is
selected at random, what is the (empirical) probability that
(A) He or she has tried the diet cola or the regular cola? What are the (empirical)
odds for this event?
(B) He or she has tried one of the colas but not both? What are the (empirical) odds
against this event?
SOLUTION Let D be the event that a person has tried the diet cola and R the
event that a person has tried the regular cola. The events D and R can be used to
partition the population of Springfield into four mutually exclusive subsets (a
collection of subsets is mutually exclusive if the intersection of any two of them
is the empty set):
D ¨ R = set of people who have tried both colas

D ¨ R′ = set of people who have tried the diet cola but not the regular cola
D′ ¨ R = set of people who have tried the regular cola but not the diet cola
D′ ¨ R′ = set of people who have not tried either cola
These sets are displayed in the Venn diagram in Figure 7.

The sample population of 1,000 persons is also partitioned into four mutually
exclusive sets, with n1D2 = 500, n1R2 = 600, and n1D ¨ R2 = 200. By using
a Venn diagram (Fig. 8), we can determine the number of sample points in the sets
D ¨ R′, D′ ¨ R, and D′ ¨ R′ (see Example 2, Section 6.3).
D R D R
D > R9 D > R D9 > R 300 200 400
D9 > R9 100
Figure 7 Total population Figure 8 Sample population
These numbers are displayed in a table:
Regular R No Regular R′ Totals

Diet D 200 300 500
No Diet D∙ 400 100 500
Totals 600 400 1,000
Assuming that each sample point is equally likely, we form a probability table
by dividing each entry in this table by 1,000, the total number surveyed. These are
empirical probabilities for the sample population, which we can use to approximate
the corresponding probabilities for the total population.
Regular R No Regular R′ Totals

Diet D .2 .3 .5
No Diet D∙ .4 .1 .5
Totals .6 .4 1.0
Now we are ready to compute the required probabilities.

(A) The event that a person has tried the diet cola or the regular cola is E = D ∪ R.
We compute P1E2 two ways:
Method 1. Directly:
P1E2 = P1D ∪ R2
= P1D2 + P1R2 - P1D ¨ R2
= .5 + .6 - .2 = .9
Method 2. Using the complement of E:
P1E2 = 1 - P1E′2
= 1 - P1D′ ¨ R′2 E′ = 1D ∪ R2 ′ = D′ ¨ R′ (see Fig. 8)
= 1 - .1 = .9
In either case,
P1E2 .9 9
odds for E = = = or 9 : 1
P1E′2 .1 1
(B) The event that a person has tried one cola but not both is the event that the per-
son has tried diet cola and not regular cola or has tried regular cola and not diet
cola. In terms of sets, this is event E = 1D ¨ R′2 ∪ 1D′ ¨ R2. Since D ¨ R′
and D′ ¨ R are mutually exclusive (Fig. 6),
P1E2 = P31D ¨ R′2 ∪ 1D′ ¨ R24
= P1D ¨ R′2 + P1D′ ¨ R2
= .3 + .4 = .7
P1E′2 .3 3
odds against E = = = or 3 : 7
P1E2 .7 7
Matched Problem 8 Refer to Example 8. If a person from Springfield is se-

lected at random, what is the (empirical) probability that
(A) He or she has not tried either cola? What are the (empirical) odds for this
event?
(B) He or she has tried the diet cola or has not tried the regular cola? What are the
(empirical) odds against this event?
Exercises 7.2
Skills Warm-up Exercises In a lottery game, a single ball is drawn at random from a con-
tainer that contains 25 identical balls numbered from 1 through
W In Problems 1–6, write the expression as a quotient of integers, 25. In Problems 25–32, use equation (1) to compute the probabil-
reduced to lowest terms. (If necessary, review Section B.1). ity that the number drawn is
3 5 1
25. Odd or a multiple of 4
10 12 8
1. 2. 3. 26. Even or a multiple of 3
9 7 3
10 12 7 27. Prime or greater than 20
4 2 4 28. Less than 10 or greater than 10
5 9 19 29. A multiple of 2 or a multiple of 5
4. 5. 6.
5 2 4
1 - 1 - 30. A multiple of 3 or a multiple of 4
6 9 19
31. Less than 5 or greater than 20
A Problems 7–12 refer to the Venn diagram below for events A and
32. Prime or less than 10
B in an equally likely sample space S. Find each of the indicated
probabilities. 33. If the probability is .51 that a candidate wins the election,
S what is the probability that he loses?
A B 34. If the probability is .03 that an automobile tire fails in less
than 50,000 miles, what is the probability that the tire does
12 38 23 not fail in 50,000 miles?
27 In Problems 35–38, use the equally likely sample space in

Example 2 to compute the probability of the following events:
7. P1A ¨ B2 8. P1A ∪ B2 35. A sum that is less than or equal to 5
9. P1A ∪ B′2 10. P1A ¨ B′2 36. A sum that is greater than or equal to 8
11. P1 1A ∪ B2 ′2 12. P1 1A ¨ B2 ′2 37. The number on the first die is a 6 or the number on the
second die is a 3.
A single card is drawn from a standard 52-card deck. Let D be
the event that the card drawn is a diamond, and let F be the event 38. The number on the first die is even or the number on the
that the card drawn is a face card. In Problems 13–24, find the second die is even.
indicated probabilities. 39. Given the following probabilities for an event E, find the
13. P1D2 14. P1F2 odds for and against E:
3 1
15. P1F′2 16. P1D′2 (A) (B) (C) .4 (D) .55
8 4
17. P1D ¨ F2 18. P1D′ ¨ F2 40. Given the following probabilities for an event E, find the
19. P1D ∪ F2 20. P1D′ ∪ F2 odds for and against E:
3 1
21. P1D ¨ F′2 22. P1D′ ¨ F′2 (A) (B) (C) .6 (D) .35
5 7
23. P1D ∪ F′2 24. P1D′ ∪ F′2
41. Compute the probability of event E if the odds in favor of Use these frequencies to calculate the approximate empirical
E are probabilities and odds for the events in Problems 59 and 60.
3 11 4 49 59. (A) The sum is less than 4 or greater than 9.

(A) (B) (C) (D)
8 7 1 51 (B) The sum is even or exactly divisible by 5.
42. Compute the probability of event E if the odds in favor of 60. (A) The sum is a prime number or is exactly divisible by 4.
E are (B) The sum is an odd number or exactly divisible by 3.
5 4 3 23 In Problems 61–64, a single card is drawn from a standard
(A) (B) (C) (D)
9 3 7 77 52-card deck. Calculate the probability of each event.
61. A face card or a club is drawn.
statement is always true, explain why. If not, give a counterexample. 62. A king or a heart is drawn.
1 63. A black card or an ace is drawn.
43. If the odds for E equal the odds against E′, then P1E2 = .
2
64. A heart or a number less than 7 (count an ace as 1) is drawn.
44. If the odds for E are a: b, then the odds against E are b: a. 65. What is the probability of getting at least 1 diamond in a
45. If P1E2 + P1F2 = P1E ∪ F2 + P1E ¨ F2, then E and F 5-card hand dealt from a standard 52-card deck?
are mutually exclusive events. 66. What is the probability of getting at least 1 face card in a
46. The theoretical probability of an event is less than or equal to 6-card hand dealt from a standard 52-card deck?
its empirical probability. 67. What is the probability that a number selected at random from
47. If E and F are complementary events, then E and F are mutu- the first 100 positive integers is (exactly) divisible by 6 or 8?
ally exclusive. 68. What is the probability that a number selected at random from
48. If E and F are mutually exclusive events, then E and F are the first 200 positive integers is (exactly) divisible by 4 or 10?
complementary. C 69. Explain how the three events A, B, and C from a sample
space S are related to each other in order for the following
B In Problems 49–52, compute the odds in favor of obtaining
equation to hold true:
49. A head in a single toss of a coin
P1A ∪ B ∪ C2 = P1A2 + P1B2 + P1C2 - P1A ¨ B2
50. A number divisible by 3 in a single roll of a die
70. Explain how the three events A, B, and C from a sample
51. At least 1 head when a single coin is tossed 3 times space S are related to each other in order for the following
52. 1 head when a single coin is tossed twice equation to hold true:
P1A ∪ B ∪ C2 = P1A2 + P1B2 + P1C2
In Problems 53–56, compute the odds against obtaining
71. Show that the solution to the birthday problem in Example 5
53. A number lesser than 5 in a single roll of a die can be written in the form
54. 2 heads when a single coin is tossed twice
365Pn
55. A 3 or an even number in a single roll of a die P1E2 = 1 - n
365
56. An odd number or a number divisible by 3 in a single roll of a die
For a calculator that has a nPr function, explain why this form
57. (A) What are the odds for rolling a sum of 5 in a single roll may be better for direct evaluation than the other form used
of two fair dice? in the solution to Example 5. Try direct evaluation of both
forms on a calculator for n = 25.
(B) If you bet $1 that a sum of 5 will turn up, what should
the house pay (plus returning your $1 bet) if a sum of 72. Many (but not all) calculators experience an overflow error
5 turns up in order for the game to be fair? when computing 365Pn for n 7 39 and when computing 365n.
Explain how you would evaluate P1E2 for any n 7 39 on
58. (A) What are the odds for rolling a sum of 10 in a single roll
such a calculator.
of two fair dice?
73. In a group of n people 1n … 122, what is the probability that
(B) If you bet $1 that a sum of 10 will turn up, what should
at least 2 of them have the same birth month? (Assume that
the house pay (plus returning your $1 bet) if a sum of
any birth month is as likely as any other.)
10 turns up in order for the game to be fair?
74. In a group of n people 1n … 1002, each person is asked to
A pair of dice are rolled 1,000 times with the following frequen-
select a number between 1 and 100, write the number on a
cies of outcomes:
slip of paper and place the slip in a hat. What is the probabil-
Sum 2 3 4 5 6 7 8 9 10 11 12 ity that at least 2 of the slips in the hat have the same number
Frequency 10 30 50 70 110 150 170 140 120 80 70 written on them?
75. If the odds in favor of an event E occurring are a to b, show that

a 12
P1E2 =
a + b
[Hint: Solve the equation P1E2 >P1E′2 = a>b for P1E2.]
76. If P1E2 = c>d, show that odds in favor of E occurring are
c to d - c.
77. The command in Figure A was used on a graphing calculator 24 0 14
to simulate 50 repetitions of rolling a pair of dice and recording
their sum. A statistical plot of the results is shown in Figure B. (B)
(A) Use Figure B to find the empirical probability of rolling
a 7 or 8.
(B) What is the theoretical probability of rolling a 7 or 8?
Applications
(C) Use a graphing calculator to simulate 200 repetitions of
rolling a pair of dice and recording their sum, and find 79. Market research. From a survey involving 1,000 university
the empirical probability of rolling a 7 or 8. students, a market research company found that 750 students
owned laptops, 450 owned cars, and 350 owned cars and
laptops. If a university student is selected at random, what is
the (empirical) probability that
(A) The student owns either a car or a laptop?
(B) The student owns neither a car nor a laptop?
80. Market research. Refer to Problem 79. If a university student
(A) is selected at random, what is the (empirical) probability that
(A) The student does not own a car?
(B) The student owns a car but not a laptop?
14
81. Insurance. By examining the past driving records of city
drivers, an insurance company has determined the following
(empirical) probabilities:
Miles Driven per Year

Less 10,000– More
Than 15,000, Than
24 0 14
10,000, Inclusive, 15,000,
M1 M2 M3 Totals
(B)
Accident A .05 .1 .15 .3
78. Consider the command in Figure A and the associated statis- No Accident A∙ .15 .2 .35 .7
tical plot in Figure B. Totals .2 .3 .5 1.0
(A) Explain why the command does not simulate 50 repeti-
tions of rolling a pair of dice and recording their sum. If a city driver is selected at random, what is the probability that
(A) He or she drives less than 10,000 miles per year or has
(B) Describe an experiment that is simulated by this command.
an accident?
(C) Simulate 200 repetitions of the experiment you described (B) He or she drives 10,000 or more miles per year and has
in part (B). Find the empirical probability of recording a no accidents?
7 or 8, and the theoretical probability of recording a 7 or 8.
82. Insurance. Use the (empirical) probabilities in Problem 81
to find the probability that a city driver selected at random
(A) Drives more than 15,000 miles per year or has an
accident
(B) Drives 15,000 or fewer miles per year and has an
accident
83. Quality control. A shipment of 60 game systems, includ-
ing 9 that are defective, is sent to a retail store. The receiving
department selects 10 at random for testing and rejects the
(A) whole shipment if 1 or more in the sample are found to be
SECTION 7.3 Conditional Probability, Intersection, and Independence 399
defective. What is the probability that the shipment will be 87. Politics. If a London resident is selected at random, what is
rejected? the (empirical) probability that the resident is
84. Quality control. An assembly plant produces 40 outboard (A) A member of the Labor party or has no preference?
motors, including 7 that are defective. The quality control What are the odds against this event?
department selects 10 at random (from the 40 produced) for
testing and will shut down the plant for troubleshooting if 1 (B) Not a member of the Labor party and prefers candidate
or more in the sample are found to be defective. What is the B? What are the odds for this event?
probability that the plant will be shut down? 88. Politics. If a London resident is selected at random, what is
85. Medicine. In order to test a new drug for adverse reactions, the the (empirical) probability that the resident is
drug was administered to 1,000 test subjects with the following (A) A member of the Conservative party or prefers candidate
results: 60 subjects reported that their only adverse reaction was A? What are the odds for this event?
a loss of appetite, 90 subjects reported that their only adverse
reaction was a loss of sleep, and 800 subjects reported no (B) Not a member of the Conservative party and has no pref-
adverse reactions at all. If this drug is released for general use, erence? What are the odds against this event?
what is the (empirical) probability that a person using the drug
will suffer both a loss of appetite and a loss of sleep? Answers to Matched Problems
86. Product testing. To test a new car, an automobile manufac- 1 2 1 7
1. (A) (B) 2. (A) (B)
turer wants to select 4 employees to test-drive the car for 1 3 3 12 18
year. If 12 management and 8 union employees volunteer to 47
be test drivers and the selection is made at random, what is 3. ≈ .336 4. .92 5. .016
140
the probability that at least 1 union employee is selected? 5
6. (A) 5 : 31 (B) $31 7. ≈ .455
Problems 87 and 88 refer to the data in the following table, 11
obtained from a random survey of 1,000 residents of London. The 1
participants were asked their political affiliations and their prefer- 8. (A) P1D′ ¨ R′2 = .1; odds for D′ ¨ R′ = or 1 : 9
9
ences in the second round of the London mayoral election. (In the 2
table, C = Conservative, L = Labor, and O = Other.) (B) P1D ∪ R′2 = .6; odds against D ∪ R′ = or 2 : 3
3
C L O Totals
Candidate A A 250 120 30 400
Candidate B B 30 400 20 450
No Preference N 20 30 100 150
Totals 300 550 150 1,000
7.3 Conditional Probability, Intersection, and Independence

■■ Conditional Probability In Section 7.2, we learned that the probability of the union of two events is related to
■■ Intersection of Events: Product Rule
the sum of the probabilities of the individual events (Theorem 1, p. 388):
■■ Probability Trees P1A ∪ B2 = P1A2 + P1B2 - P1A ¨ B2
■■ Independent Events In this section, we will learn how the probability of the intersection of two events is
related to the product of the probabilities of the individual events. But first we must
■■ Summary
investigate the concept of conditional probability.
Conditional Probability
The probability of an event may change if we are told of the occurrence of another
event. For example, if an adult (21 years or older) is selected at random from all
adults in the United States, the probability of that person having lung cancer would
not be high. However, if we are told that the person is also a heavy smoker, we would
want to revise the probability upward.
In general, the probability of the occurrence of an event A, given the occurrence
of another event B, is called a conditional probability and is denoted by P1A∙ B2.
In the preceding situation, events A and B would be

A = adult has lung cancer
B = adult is a heavy smoker
and P1A∙ B2 would represent the probability of an adult having lung cancer, given
that he or she is a heavy smoker.
1 A 4 Our objective now is to try to formulate a precise definition of P1A∙ B2. It is
3 6 helpful to start with a relatively simple problem, solve it intuitively, and then general-
2 5 ize from this experience.
What is the probability of rolling a prime number (2, 3, or 5) in a single roll of a
fair die? Let
Figure 1
S = 51, 2, 3, 4, 5, 66
Then the event of rolling a prime number is (Fig. 1)

A = 52, 3, 56
Thus, since we assume that each simple event in the sample space is equally likely,
n1A2 3 1
B P1A2 = = =
n1S2 6 2
1 A 4
3 6 Now suppose you are asked, “In a single roll of a fair die, what is the probability
2 5
that a prime number has turned up if we are given the additional information that
an odd number has turned up?” The additional knowledge that another event has
occurred, namely,
B = odd number turns up
1
puts the problem in a new light. We are now interested only in the part of event A
3
A>B (rolling a prime number) that is in event B (rolling an odd number). Event B, since we
5 know it has occurred, becomes the new sample space. The Venn diagrams in Figure 2
illustrate the various relationships. Thus, the probability of A given B is the number of
Figure 2 B is the new
A elements in B divided by the total number of elements in B. Symbolically,
sample space.
n1A ¨ B2 2
P1A∙ B2 = =
n1B2 3
Dividing the numerator and denominator of n1A ¨ B2 >n1B2 by n1S2, the number of
elements in the original sample space, we can express P1A∙ B2 in terms of P1A ¨ B2
and P1B2:*
n1A ¨ B2
n1A ¨ B2 n1S2 P1A ¨ B2
P1A∙ B2 = = =
n1B2 n1B2 P1B2
n1S2
Using the right side to compute P1A∙ B2 for the preceding example, we obtain the
same result:
2
P1A ¨ B2 6 2
P1A∙ B2 = = 3
=
P1B2 6
3
We use the formula above to motivate the following definition of conditional

probability, which applies to any sample space, including those having simple events
that are not equally likely (see Example 1).
*Note that P1A∙ B2 is a probability based on the new sample space B, while P1A ¨ B2 and P1B2 are both
probabilities based on the original sample space S.
DEFINITION Conditional Probability

For events A and B in an arbitrary sample space S, we define the conditional
probability of A given B by
P1A ¨ B2
P1A∙ B2 = P1B2 ∙ 0 (1)
P1B2
EXAMPLE 1 Conditional Probability A pointer is spun once on a circular spinner (Fig. 3). The
probability assigned to the pointer landing on a given integer (from 1 to 6) is the
ratio of the area of the corresponding circular sector to the area of the whole circle,
as given in the table:
ei 1 2 3 4 5 6
S = 51, 2, 3, 4, 5, 66
P1ei 2 .1 .2 .1 .1 .3 .2
(A) What is the probability of the pointer landing on a prime number?

(B) What is the probability of the pointer landing on a prime number, given that it
landed on an odd number?
SOLUTION Let the events E and F be defined as follows:
E = pointer lands on a prime number = 52, 3, 56

Figure 3
F = pointer lands on an odd number = 51, 3, 56
(A) P1E2 = P122 + P132 + P152
= .2 + .1 + .3 = .6
(B) First note that E ¨ F = 53, 56.
P1E ¨ F2 P132 + P152
P1E ∙ F2 = =
P1F2 P112 + P132 + P152
.1 + .3 .4
= = = .8
.1 + .1 + .3 .5
(A) What is the probability of the pointer landing on a number greater than 4?
(B) What is the probability of the pointer landing on a number greater than 4,
given that it landed on an even number?
EXAMPLE 2 Safety Research Suppose that city records produced the following probability
data on a driver being in an accident on the last day of a Memorial Day weekend:
Accident A No Accident A′ Totals

Rain R .025 .335 .360
No Rain R∙ .015 .625 .640 S = 5RA, RA′, R′A, R′A′6
Totals .040 .960 1.000
(A) Find the probability of an accident, rain or no rain.

(B) Find the probability of rain, accident or no accident.
(C) Find the probability of an accident and rain.
(D) Find the probability of an accident, given rain.
SOLUTION
(A) Let A = 5RA, R′A6 Event: “accident”
P1A2 = P1RA2 + P1R′A2 = .025 + .015 = .040
(B) Let R = 5RA, RA′6 Event: “rain”
P1R2 = P1RA2 + P1RA′2 = .025 + .335 = .360
(C) A ¨ R = 5RA6 Event: “accident and rain”
P1A ¨ R2 = P1RA2 = .025
P1A ¨ R2 .025
(D) P1A∙ R2 = = = .069 Event: “accident, given rain”
P1R2 .360
Compare the result in part (D) with that in part (A). Note that P1A∙ R2 ∙ P1A2,
and the probability of an accident, given rain, is higher than the probability of an
accident without the knowledge of rain.
Matched Problem 2 Referring to the table in Example 2, determine the

following:
(A) Probability of no rain
(B) Probability of an accident and no rain
(C) Probability of an accident, given no rain [Use formula (1) and the results of
parts (A) and (B).]
Intersection of Events: Product Rule

Let’s return to the original problem of this section, that is, representing the prob-
ability of an intersection of two events in terms of the probabilities of the individual
events. If P1A2 ∙ 0 and P1B2 ∙ 0, then using formula (1), we can write
P1A ¨ B2 P1B ¨ A2
P1A∙ B2 = and P1B∙ A2 =
P1B2 P1A2
Solving the first equation for P1A ¨ B2 and the second equation for P1B ¨ A2, we
have
P1A ¨ B2 = P1B2P1A∙ B2 and P1B ¨ A2 = P1A2P1B∙ A2
Since A ¨ B = B ¨ A for any sets A and B, it follows that
P1A ¨ B2 = P1B2P1A∙ B2 = P1A2P1B∙ A2
and we have the product rule:
THEOREM 1 Product Rule

For events A and B with nonzero probabilities in a sample space S,
P1A ¨ B2 = P1A2P1B∙ A2 = P1B2P1A∙ B2 (2)
and we can use either P1A2P1B∙ A2 or P1B2P1A∙ B2 to compute P1A ¨ B2.
EXAMPLE 3 Consumer Survey If 60% of a department store’s customers are female and 75%
of the female customers have credit cards at the store, what is the probability that a
customer selected at random is a female and has a store credit card?
SOLUTION Let
S = all store customers

F = female customers
C = customers with a store credit card
If 60% of the customers are female, then the probability that a customer selected at
random is a female is
P1F2 = .60
Since 75% of the female customers have store credit cards, the probability that a
customer has a store credit card, given that the customer is a female, is
P1C∙ F2 = .75
Using equation (2), the probability that a customer is a female and has a store credit
card is
P1F ¨ C2 = P1F2P1C∙ F2 = 1.6021.752 = .45
Matched Problem 3 If 80% of the male customers of the department store in

Example 3 have store credit cards, what is the probability that a customer selected
at random is a male and has a store credit card?
Probability Trees
We used tree diagrams in Section 6.3 to help us count the number of combined out-
comes in a sequence of experiments. In a similar way, we will use probability trees to
Figure 4 help us compute the probabilities of combined outcomes in a sequence of experiments.
EXAMPLE 4 Probability Tree Two balls are drawn in succession, without replacement, from a
box containing 3 blue and 2 white balls (Fig. 4). What is the probability of drawing
a white ball on the second draw?
SOLUTION We start with a tree diagram (Fig. 5) showing the combined outcomes
FIRST SECOND of the two experiments (first draw and second draw). Then we assign a probability
DRAW DRAW to each branch of the tree (Fig. 6). For example, we assign the probability 25 to the
2 branch Sw1, since this is the probability of drawing a white ball on the first draw
1
S
b2 (there are 2 white balls and 3 blue balls in the box). What probability should be
Start b1
2 assigned to the branch w1w2? This is the conditional probability P1w2 ∙ w1 2, that is,
b2 the probability of drawing a white ball on the second draw given that a white ball
Figure 5 was drawn on the first draw and not replaced. Since the box now contains 1 white
ball and 3 blue balls, the probability is 14. Continuing in the same way, we assign
FIRST SECOND probabilities to the other branches of the tree and obtain Figure 6.
DRAW DRAW What is the probability of the combined outcome w1 ¨ w2, that is, the probabil-
1
2
4 2 ity of drawing a white ball on the first draw and a white ball on the second draw?*
2
2 1 Using the product rule (2), we have
5
3
2 b2
S
4 P1w1 ¨ w2 2 = P1w1 2P1w2 ∙ w1 2
2
2
4 2
3
2 2 1 1
5 b1 = a ba b =
2
2 b2
5 4 10
4
Figure 6 The combined outcome w1 ¨ w2 corresponds to the unique path Sw1w2 in the tree
diagram, and we see that the probability of reaching w2 along this path is the product
of the probabilities assigned to the branches on the path. Reasoning in this way, we
*
The sample space for the combined outcomes is S = 5w1w2, w1b2, b1w2, b1b2 6. If we let
w1 = 5w1w2, w1b2 6 and w2 = 5w1w2, b1w2 6, then w1 ¨ w2 = 5w1w2 6.
obtain the probability of each remaining combined outcome by multiplying the prob-
abilities assigned to the branches on the path corresponding to the given combined
outcomes. These probabilities are often written at the ends of the paths to which they
correspond (Fig. 7).
FIRST SECOND
DRAW DRAW
1 1 2 1 1
2
4 2 2
10
2
5
?2
4
52
10
2 1
2
5
3 3 2 3 3
2
4
b2 2
10
2
5
?2 2
4 5 10
S 2 3 3 2 3
2
4 2 2
10
2
5
?2
4
52
10
3
2
5 b1
2 3 3 2 3
2
4
b2 2
10
2
5
?2
4
52
10
Figure 7 Sum 5 1
Now we can complete the problem. A white ball drawn on the second draw
corresponds to either the combined outcome w1 ¨ w2 or b1 ¨ w2 occurring. Thus,
since these combined outcomes are mutually exclusive,
P1w2 2 = P1w1 ¨ w2 2 + P1b1 ¨ w2 2
1 3 4 2
= + = =
10 10 10 5
which is the sum of the probabilities listed at the ends of the two paths terminating
in w2.
Matched Problem 4 Two balls are drawn in succession without replacement

from a box containing 4 red and 2 white balls. What is the probability of drawing
a red ball on the second draw?
The sequence of two experiments in Example 4 is an example of a stochastic

process. In general, a stochastic process involves a sequence of experiments where
the outcome of each experiment is not certain. Our interest is in making predictions
about the process as a whole. The analysis in Example 4 generalizes to stochastic
processes involving any finite sequence of experiments. We summarize the proce-
dures used in Example 4 for general application:
PROCEDURE Constructing Probability Trees
Step 1 Draw a tree diagram corresponding to all combined outcomes of the

sequence of experiments.
Step 2 Assign a probability to each tree branch. (This is the probability of the occur-
rence of the event on the right end of the branch subject to the occurrence of
all events on the path leading to the event on the right end of the branch. The
probability of the occurrence of a combined outcome that corresponds to a
path through the tree is the product of all branch probabilities on the path.*)
Step 3 Use the results in Steps 1 and 2 to answer various questions related to the
sequence of experiments as a whole.
*If we form a sample space S such that each simple event in S corresponds to one path through the tree,
and if the probability assigned to each simple event in S is the product of the branch probabilities on the
corresponding path, then it can be shown that this is not only an acceptable assignment (all probabilities
for the simple events in S are nonnegative and their sum is 1), but it is the only assignment consistent
with the method used to assign branch probabilities within the tree.

Refer to the table on rain and accidents in Example 2 and use formula (1), where
appropriate, to complete the following probability tree:
A .025
? A .025 RA
.360 R
.360 R
A9 ? A9 ? RA9
S S
A ? A ? R9A
R9 ?
R9
A9 ? A9 ? R9A9
Discuss the difference between P1R ¨ A2 and P1A∙ R2.
EXAMPLE 5 Product Defects An auto company A subcontracts the manufacturing of its

onboard computers to two companies: 40% to company B and 60% to company C.
Company B in turn subcontracts 70% of the orders it receives from company A to
company D and the remaining 30% to company E, both subsidiaries of company
B. When the onboard computers are completed by companies D, E, and C, they
are shipped to company A to be used in various car models. It has been found that
1.5%, 1%, and .5% of the boards from D, E, and C, respectively, prove defective
during the 3-year warranty period after a car is first sold. What is the probability
that a given onboard computer will be defective during the 3-year warranty period?
SOLUTION Draw a tree diagram and assign probabilities to each branch (Fig. 8):
.985 OK
D
.7
.015 Defective
B
.4 .99 OK
.3
E
A .01 Defective
.6 .995 OK
C
Figure 8 .005 Defective
There are three paths leading to defective (the onboard computer will be defec-
tive within the 3-year warranty period). We multiply the branch probabilities on
each path and add the three products:
P1defective2 = 1.421.721.0152 + 1.421.321.012 + 1.621.0052
= .0084
Matched Problem 5 In Example 5, what is the probability that a given on-

board computer came from company E or C?
Independent Events
We return to Example 4, which involved drawing two balls in succession without
replacement from a box of 3 blue and 2 white balls. What difference does “without
replacement” and “with replacement” make? Figure 9 shows probability trees corre-
sponding to each case. Go over the probability assignments for the branches in Figure
9B to convince yourself of their correctness.
FIRST SECOND COMBINED FIRST SECOND COMBINED

DRAW DRAW OUTCOMES DRAW DRAW OUTCOMES
1 2
2
4 2 .10 1 2 2
5 2 .16 1 2
2 1 2 1
2
5
2
5
3 3
2
4
b2 .30 1 b2 2
5
b2 .24 1 b2
S 2 S 2
2
4 2 .30 b1 2 2
5 2 .24 b1 2
3 3
2
5 b1 2
5 b1
2 3
2
4
b2 .30 b1b2 2
5
b2 .36 b1b2
(A) Without replacement (B) With replacement
Figure 9 S = 5w1w2, w1b2, b1w2, b1b2 6
Let
A = white ball on second draw = 5w1w2, b1w2 6
B = white ball on first draw = 5w1w2, w1b2 6
We now compute P1A∙ B2 and P1A2 for each case in Figure 9.
Case 1. Without replacement:
P1A ¨ B2 P5w1w2 6 .10

P1A ∙ B2 = = = = .25
P1B2 P5w1w2, w1b2 6 .10 + .30
(This is the assignment to branch w1w2 that we made by looking in the box and
counting.)
P1A2 = P5w1w2, b1w2 6 = .10 + .30 = .40
Note that P1A∙ B2 ∙ P1A2, and we conclude that the probability of A is affected by
the occurrence of B.
Case 2. With replacement:
P1A ¨ B2 P5w1w2 6 .16

P1A ∙ B2 = = = = .40
P1B2 P5w1w2, w1b2 6 .16 + .24
(This is the assignment to branch w1w2 that we made by looking in the box and
counting.)
P1A2 = P5w1w2, b1w2 6 = .16 + .24 = .40
Note that P1A∙ B2 = P1A2, and we conclude that the probability of A is not affected
by the occurrence of B.
Intuitively, if P1A∙ B2 = P1A2, then it appears that event A is “independent” of
B. Let us pursue this further. If events A and B are such that
P1A∙ B2 = P1A2
then replacing the left side by its equivalent from formula (1), we obtain
P1A ¨ B2
= P1A2
P1B2
After multiplying both sides by P1B2, the last equation becomes
P1A ¨ B2 = P1A2P1B2
This result motivates the following definition of independence:
DEFINITION Independence
If A and B are any events in a sample space S, we say that A and B are indepen-
dent if
P1A ¨ B2 = P1A2P1B2 (3)
Otherwise, A and B are said to be dependent.
From the definition of independence one can prove (see Problems 75 and 76,
Exercises 7.3) the following theorem:
THEOREM 2 On Independence
If A and B are independent events with nonzero probabilities in a sample space
S, then
P1A∙ B2 = P1A2 and P1B∙ A2 = P1B2 (4)
If either equation in (4) holds, then A and B are independent.
CONCEPTUAL INSIGHT
Sometimes intuitive reasoning can be helpful in deciding whether or not two
events are independent. Suppose that a fair coin is tossed five times. What is the
probability of a head on the fifth toss, given that the first four tosses are all heads?
Our intuition tells us that a coin has no memory, so the probability of a head on
the fifth toss given four previous heads should be equal to the probability of a
head on the fifth toss, namely, 1/2. In other words, the first equation of Theorem
2 holds intuitively, so “heads on the fifth toss” and “heads on the first four tosses”
are independent events.
Often, unfortunately, intuition is not a reliable guide to the notion of indepen-
dence. Independence is a technical concept. So in all cases, an appropriate sample
space should be chosen, and either equation (3) or equation (4) should be tested,
to confirm that two events are (or are not) independent.
EXAMPLE 6 Testing for Independence In two tosses of a single fair coin, show that the events
“A head on the first toss” and “A head on the second toss” are independent.
SOLUTION Consider the sample space of equally likely outcomes for the tossing of
a fair coin twice,
S = 5HH, HT, TH, TT6
and the two events,
A = a head on the first toss = 5HH, HT6
B = a head on the second toss = 5HH, TH6
Then
2 1 2 1 1
P1A2 = = P1B2 = = P1A ¨ B2 =
4 2 4 2 4
Thus,
= # = P1A2P1B2
1 1 1
P1A ¨ B2 =
4 2 2
and the two events are independent. (The theory agrees with our intuition—a coin
has no memory.)
Matched Problem 6 In Example 6, compute P1B∙ A2 and compare with P1B2.
EXAMPLE 7 Testing for Independence A single card is drawn from a standard 52-card deck.
Test the following events for independence (try guessing the answer to each part
before looking at the solution):
(A) E = the drawn card is a spade.
F = the drawn card is a face card.
(B) G = the drawn card is a club.
H = the drawn card is a heart.
SOLUTION
(A) To test E and F for independence, we compute P1E ¨ F2 and P1E2P1F2. If
they are equal, then events E and F are independent; if they are not equal, then
events E and F are dependent.
3 13 12 3
P1E ¨ F2 = P1E2P1F2 = a ba b =
52 52 52 52
Events E and F are independent. (Did you guess this?)
(B) Proceeding as in part (A), we see that
13 13 1
P1G ¨ H2 = P1∅2 = 0 P1G2P1H2 = a ba b =
52 52 16
Events G and H are dependent. (Did you guess this?)
Students often confuse mutually exclusive (disjoint) events with
! CAUTION
independent events. One does not necessarily imply the other.
In fact, it is not difficult to show (see Problem 79, Exercises 7.3) that any two mutu-
ally exclusive events A and B, with nonzero probabilities, are always dependent.
Matched Problem 7 A single card is drawn from a standard 52-card deck.

Test the following events for independence:
(A) E = the drawn card is a red card
F = the drawn card’s number is divisible by 5 1face cards are not assigned
values)
(B) G = the drawn card is a king
H = the drawn card is a queen

In soccer, would it be reasonable to assume that the following events are indepen-
dent? Explain why or why not.
A = the Manchester United FC win in the first round of the Premier League.
B = the Manchester United FC win in the second round of the Premier League.
The notion of independence can be extended to more than two events:

DEFINITION Independent Set of Events

A set of events is said to be independent if for each finite subset 5E1, E2, c, Ek 6
P1E1 ¨ E2 ¨ g ¨ Ek 2 = P1E1 2P1E2 2 # g # P1Ek 2 (5)
The next example makes direct use of this definition.
EXAMPLE 8 Computer Control Systems A space shuttle has four independent computer con-
trol systems. If the probability of failure (during flight) of any one system is .001,
what is the probability of failure of all four systems?
SOLUTION Let
E1 = failure of system 1 E3 = failure of system 3

E2 = failure of system 2 E4 = failure of system 4
Then, since events E1, E2, E3, and E4 are given to be independent,
P1E1 ¨ E2 ¨ E3 ¨ E4 2 = P1E1 2P1E2 2P1E3 2P1E4 2
= 1.0012 4
= .000 000 000 001
Matched Problem 8 A single die is rolled 6 times. What is the probability of

getting the sequence 1, 2, 3, 4, 5, 6?
Summary
The key results in this section are summarized in the following box:
SUMMARY Key Concepts

P1A ¨ B2 P1B ¨ A2
P1A∙ B2 = P1B∙ A2 =
P1B2 P1A2
Note: P1A∙ B2 is a probability based on the new sample space B, while P1A ¨ B2
and P1B2 are probabilities based on the original sample space S.
Product Rule
P1A ¨ B2 = P1A2P1B∙ A2 = P1B2P1A∙ B2
Independent Events
• A and B are independent if
P1A ¨ B2 = P1A2P1B2
• If A and B are independent events with nonzero probabilities, then
P1A∙ B2 = P1A2 and P1B∙ A2 = P1B2
• If A and B are events with nonzero probabilities and either P1A∙ B2 = P1A2
or P1B∙ A2 = P1B2, then A and B are independent.
• If E1, E2, c, En are independent, then
P1E1 ¨ E2 ¨ g ¨ En 2 = P1E1 2P1E2 2 # g # P1En 2
Exercises 7.3
Skills Warm-up Exercises In Problems 23–26, find each probability directly from the table.
W In Problems 1–6, use a tree diagram to represent a factorization 23. P1B2 24. P1E2
of the given integer into primes, so that there are two branches 25. P1B ¨ D2 26. P1C ¨ E2
at each number that is not prime. For example, the factorization
24 = 4 # 6 = 12 # 22 # 12 # 32 is represented by: In Problems 27–34, compute each probability using formula (1)
on page 401 and appropriate table values.
2
4
2 27. P1D∙ B2 28. P1C ∙ E2
24
2
6
3
29. P1B ∙ D2 30. P1E ∙ C2
31. P1C ∙ F2 32. P1F ∙ C2
(If necessary, review Section B.3.)
33. P1A ∙ C2 34. P1B ∙ B2
1. 100 2. 120
3. 180 4. 225 In Problems 35–42, test each pair of events for independence.
5. 315 6. 360 35. A and D 36. A and E

37. B and D 38. B and E
A A single card is drawn from a standard 52-card deck. In Problems 7–14,
find the conditional probability that 39. B and F 40. C and F
7. The card is an ace, given that it is a heart. 41. A and B 42. D and F
8. The card is a heart, given that it is a nonface card. B 43. A fair coin is tossed 8 times.
9. The card is a heart, given that it is an ace. (A) What is the probability of tossing a head on the 8th toss,
given that the preceding 7 tosses were heads?
10. The card is a nonface card, given that it is a heart.
(B) What is the probability of getting 8 heads or 8 tails?
11. The card is black, given that it is a club.
44. A fair die is rolled 5 times.
12. The card is a jack, given that it is red. (A) What is the probability of getting a 6 on the 5th roll,
13. The card is a club, given that it is black. given that a 6 turned up on the preceding 4 rolls?
14. The card is red, given that it is a jack. (B) What is the probability that the same number turns up
every time?
In Problems 15–22, find the conditional probability, in a single 45. A pointer is spun once on the circular spinner shown below.
roll of two fair dice, that The probability assigned to the pointer landing on a given
15. The sum is less than 6, given that the sum is even. integer (from 1 to 5) is the ratio of the area of the correspond-
ing circular sector to the area of the whole circle, as given in
16. The sum is 10, given that the roll is doubles. the table:
17. The sum is even, given that the sum is less than 6. ei 1 2 3 4 5
18. The roll is doubles, given that the sum is 10. P1ei 2 .3 .1 .2 .3 .1
19. Neither die is a one, given that the sum is lesser than 6.
20. The sum is odd, given that at least one die is a six.
21. The sum is lesser than 6, given that neither die is a one.
22. At least one die is a six, given that the sum is odd.
In Problems 23–42, use the table below. Events A, B, and C are

mutually exclusive; so are D, E, and F.
A B C Totals
D .20 .03 .07 .30
E .28 .05 .07 .40 Given the events
F .22 .02 .06 .30 E = pointer lands on an even number
Totals .70 .10 .20 1.00 F = pointer lands on a number less than 4
(A) Find P1F ∙ E2. (A) Find P1H∙ G2.

(B) Test events E and F for independence. (B) Test H and G for independence.
46. Repeat Problem 45 with the following events: 56. A card is drawn at random from a standard 52-card deck.
Events M and N are
E = pointer lands on an odd number
F = pointer lands on a prime number M = the drawn card is a diamond.
N = the drawn card is even 1face cards are not valued2.
Compute the indicated probabilities in Problems 47 and 48 by
(A) Find P1N ∙ M2.
referring to the following probability tree:
(B) Test M and N for independence.
.2 R
M 57. Let A be the event that all of a family’s children are the same
.3
.8 S gender, and let B be the event that the family has at most
Start
.4 R 1 boy. Assuming the probability of having a girl is the same
.7 N as the probability of having a boy (both .5), test events A and
.6 S B for independence if
(A) The family has 2 children.
47. (A) P1M ¨ S2 (B) P1R2 (B) The family has 3 children.
58. An experiment consists of tossing n coins. Let A be the event
48. (A) P1N ¨ R2 (B) P1S2 that at least 2 heads turn up, and let B be the event that all the
49. A fair coin is tossed twice. Consider the sample space coins turn up the same. Test A and B for independence if
S = 5HH, HT, TH, TT6 of equally likely simple events. (A) 2 coins are tossed.
We are interested in the following events:
(B) 3 coins are tossed.
E1 = a head on the first toss
Problems 59–62 refer to the following experiment: 2 balls are
E2 = a tail on the first toss
drawn in succession out of a box containing 2 red and 5 white
E3 = a tail on the second toss balls. Let Ri be the event that the ith ball is red, and let Wi be the
E4 = a head on the second toss event that the ith ball is white.
For each pair of events, discuss whether they are independent 59. Construct a probability tree for this experiment and find
and whether they are mutually exclusive. the probability of each of the events R1 ¨ R2, R1 ¨ W2,
(A) E1 and E4 (B) E1 and E2 W1 ¨ R2, W1 ¨ W2, given that the first ball drawn was
50. For each pair of events (see Problem 49), discuss whether they (A) Replaced before the second draw
are independent and whether they are mutually exclusive. (B) Not replaced before the second draw
(A) E1 and E3 (B) E3 and E4 60. Find the probability that the second ball was red, given that
51. In 2 throws of a fair die, what is the probability that you will the first ball was
get an even number on each throw? An even number on the (A) Replaced before the second draw
first or second throw?
(B) Not replaced before the second draw
52. In 2 throws of a fair die, what is the probability that you will get
at least 5 on each throw? At least 5 on the first or second throw? 61. Find the probability that at most 1 ball was red, given that the
first ball was
53. Two cards are drawn in succession from a standard 52-card (A) Replaced before the second draw
deck. What is the probability that the first card is a club and
the second card is a heart (B) Not replaced before the second draw
(A) If the cards are drawn without replacement? 62. Find the probability that both balls were of different colors,
given that the first ball was
(B) If the cards are drawn with replacement?
(A) Replaced before the second draw
54. Two cards are drawn in succession from a standard 52-card
deck. What is the probability that both cards are red (B) Not replaced before the second draw
(A) If the cards are drawn without replacement? C In Problems 63–70, discuss the validity of each statement. If the
(B) If the cards are drawn with replacement? statement is always true, explain why. If not, give a counterexample.
55. A card is drawn at random from a standard 52-card deck. 63. If P1A ∙ B2 = P1B2, then A and B are independent.
Events G and H are 64. If A and B are independent, then P1A ∙ B2 = P1B ∙ A2.
G = the drawn card is black. 65. If A is nonempty and A ⊂ B, then P1A ∙ B2 Ú P1A2.
H = the drawn card is divisible by 3 1face
66. If A and B are events, then P1A ∙ B2 … P1B2.
cards are not valued).
67. If A and B are mutually exclusive, then A and B are independent.
68. If A and B are independent, then A and B are mutually exclusive. Salary Classification
69. If two balls are drawn in succession, with replacement, Hourly Salary
from a box containing m red and n white balls 1m Ú 1 and (H) (S) Totals
n Ú 12, then To Yes (Y) 400 200 600
Strike No (N) 150 250 400
P1W1 ¨ R2 2 = P1R1 ¨ W2 2
Totals 550 450 1,000
70. If two balls are drawn in succession, without replacement,
from a box containing m red and n white balls 1m Ú 1 and (A) Convert this table to a probability table by dividing each
n Ú 12, then entry by 1,000.
(B) What is the probability of an employee voting to strike?
P1W1 ¨ R2 2 = P1R1 ¨ W2 2
Of voting to strike given that the person is paid hourly?
71. A box contains 2 red, 3 white, and 4 green balls. Two balls (C) What is the probability of an employee being on salary
are drawn out of the box in succession without replacement. 1S2? Of being on salary given that he or she voted in
What is the probability that both balls are the same color? favor of striking?
72. For the experiment in Problem 71, what is the probability that 82. Quality control. An automobile manufacturer produces 37%
no white balls are drawn? of its cars at plant A. If 5% of the cars manufactured at plant
73. An urn contains 2 one-dollar bills, 1 five-dollar bill, and 1 A have defective emission control devices, what is the prob-
ten-dollar bill. A player draws bills one at a time without re- ability that one of this manufacturer’s cars was manufactured
placement from the urn until a ten-dollar bill is drawn. Then at plant A and has a defective emission control device?
the game stops. All bills are kept by the player. 83. Bonus incentives. If a salesperson has gross sales of over
(A) What is the probability of winning $16? $600,000 in a year, then he or she is eligible to play the
company’s bonus game: A black box contains 1 twenty-dollar
(B) What is the probability of winning all bills in the urn?
bill, 2 five-dollar bills, and 1 one-dollar bill. Bills are drawn
(C) What is the probability of the game stopping at the out of the box one at a time without replacement until a
second draw? twenty-dollar bill is drawn. Then the game stops. The sales-
74. Ann and Barbara are playing a tennis match. The first player person’s bonus is 1,000 times the value of the bills drawn.
to win 2 sets wins the match. For any given set, the probabil- (A) What is the probability of winning a $26,000 bonus?
ity that Ann wins that set is 23. Find the probability that
(B) What is the probability of winning the maximum bonus,
(A) Ann wins the match. $31,000, by drawing out all bills in the box?
(B) 3 sets are played. (C) What is the probability of the game stopping at the third
(C) The player who wins the first set goes on to win the draw?
match. 84. Personnel selection. To transfer into a particular techni-
75. Show that if A and B are independent events with nonzero cal department, a company requires an employee to pass
probabilities in a sample space S, then a screening test. A maximum of 3 attempts are allowed at
6-month intervals between trials. From past records it is
P1A ∙ B2 = P1A2 and P1B ∙ A2 = P1B2 found that 40% pass on the first trial; of those that fail the
76. Show that if A and B are events with nonzero probabili- first trial and take the test a second time, 60% pass; and of
ties in a sample space S, and either P1A ∙ B2 = P1A2 or those that fail on the second trial and take the test a third
P1B ∙ A2 = P1B2, then events A and B are independent. time, 20% pass. For an employee wishing to transfer:
77. Show that P1A ∙ A2 = 1 when P1A2 ∙ 0. (A) What is the probability of passing the test on the first or
second try?
78. Show that P1A ∙ B2 + P1A′∙ B2 = 1.
(B) What is the probability of failing on the first 2 trials and
79. Show that A and B are dependent if A and B are mutually passing on the third?
exclusive and P1A2 ∙ 0, P1B2 ∙ 0.
(C) What is the probability of failing on all 3 attempts?
80. Show that P1A ∙ B2 = 1 if B is a subset of A and P1B2 ∙ 0. 85. European Food Safety Authority. A candy manufacturer
wishes to use a new red dye to enhance the color in its raspber-
ry ripple candies. The European Food Safety Authority (EFSA)
requires the dye to be tested for hyperactivity-producing
Applications potential using laboratory rats. The results of one test on 1,000
rats are summarized in the following table:
81. Labor relations. In a study to determine employee voting Hyperactivity No Hyperactivity
patterns in a recent strike election, 1,000 employees H H′ Totals
were selected at random and the following tabulation Ate Red Dye R 90 410 500
was made:
Did Not Eat Red 30 470 500
Dye R∙
Totals 120 880 1,000
SECTION 7.4 Bayes’ Formula 413
(A) Convert the table into a probability table by dividing (B) What is the probability of a person having an IQ below 90?
each entry by 1,000.
(C) Are events A and F dependent? A and F′?
(B) Are “developing hyperactivity” and “eating red dye”
independent events? 88. Psychology.
(C) Should the EFSA approve or ban the use of the dye? (A) What is the probability of a person having an IQ above
Explain why or why not using P1H∙ R2 and P1H2. 120, given that the person is a female? A male?
86. Genetics. In a study to determine frequency and dependency of (B) What is the probability of a person being female and
color-blindness relative to females and males, 1,000 people were having an IQ above 120?
chosen at random, and the following results were recorded:
(C) Are events C and F dependent? C and F′?
Female Male
F F′ Totals 89. Voting patterns. A survey of a Brazilian constituency’s
residents revealed that 45% of the residents were members of
Color-Blind C 2 24 26
the Workers’ party and 52% of the Workers’ party members
Normal C∙ 518 456 974 voted in the last election. What is the probability that a person
Totals 520 480 1,000 selected at random from this constituency is a member of the
Workers’ party and voted in the last election?
(A) Convert this table to a probability table by dividing each
entry by 1,000.
(B) What is the probability that a person is a woman, given
that the person is color-blind? Are the events color- 1. (A) .5 (B) .4
blindness and female independent? 2. (A) P1R′2 = .640 (B) P1A ¨ R′2 = .015
(C) What is the probability that a person is color-blind, given P1A ¨ R′2
that the person is a male? Are the events color-blindness (C) P1A ∙ R′2 = = .023
P1R′2
and male independent?
3. P1M ¨ C2 = P1M2P1C∙ M2 = .32
Problems 87 and 88 refer to the data in the following table, ob-
tained in a study to determine the frequency and dependency of IQ 2
4. 5. .72
ranges relative to males and females. 1,000 people were chosen at 3
random and the following results were recorded: 1
IQ P1A ¨ B2 4 1
6. P1B ∙ A2 = = = = P1B2
Below 90 90–120 Above 120 P1A2 1 2
(A) (B) (C) Totals 2
Female F 130 286 104 520 7. (A) E and F are independent.
Male F′ 120 264 96 480 (B) G and H are dependent.
Totals 250 550 200 1,000
1 6
8. a b ≈ .000 021 4
87. Psychology. 6
(A) What is the probability of a person having an IQ below

90, given that the person is a female? A male?
7.4 Bayes’ Formula

In the preceding section, we discussed the conditional probability of the occurrence
of an event, given the occurrence of an earlier event. Now we will reverse the prob-
lem and try to find the probability of an earlier event conditioned on the occurrence
of a later event. As you will see, a number of practical problems have this form.
First, let us consider a relatively simple problem that will provide the basis for a
generalization.
EXAMPLE 1 Probability of an Earlier Event Given a Later Event One urn has 3 blue and
2 white balls; a second urn has 1 blue and 3 white balls (Fig. 1). A single fair die is
rolled and if 1 or 2 comes up, a ball is drawn out of the first urn; otherwise, a ball is
drawn out of the second urn. If the drawn ball is blue, what is the probability that it
came out of the first urn? Out of the second urn?
Figure 1
SOLUTION We form a probability tree, letting U1 represent urn 1, U2 urn 2, B a blue

ball, and W a white ball. Then, on the various outcome branches, we assign appro-
priate probabilities. For example, P1U1 2 = 13, P1B∙ U1 2 = 35, and so on:
3
2
5 B
1
2 U1
3
2
2
5 W
Start 1
2
4 B
2
2
3 U2
3
2
4 W
Now we are interested in finding P1U1 ∙ B2, that is, the probability that the ball came
out of urn 1, given that the drawn ball is blue. Using equation (1) from Section 7.3, we
can write
P1U1 ¨ B2
P1U1 ∙ B2 = (1)
P1B2
If we look at the tree diagram, we can see that B is at the end of two different
branches; thus,
P1B2 = P1U1 ¨ B2 + P1U2 ¨ B2 (2)
After substituting equation (2) into equation (1), we get
P1U1 ¨ B2
P1U1 ∙ B2 = P1A ¨ B2 = P1A2P1B ∙ A2
P1U1 ¨ B2 + P1U2 ¨ B2
P1U1 2P1B∙ U1 2
=
P1U1 2P1B∙ U1 2 + P1U2 2P1B∙ U2 2
P1B∙ U1 2P1U1 2
= (3)
P1B∙ U1 2P1U1 2 + P1B∙ U2 2P1U2 2
Equation (3) is really a lot simpler to use than it looks. You do not need to
memorize it; you simply need to understand its form relative to the probability tree
above. Referring to the probability tree, we see that
P1B∙ U1 2P1U1 2 = product of branch probabilities leading to B through U1
3 1
= a ba b We usually start at B and work back through U1.
5 3
P1B∙ U2 2P1U2 2 = product of branch probabilities leading to B through U2
1 2
= a ba b We usually start at B and work back through U2.
4 3
Equation (3) now can be interpreted in terms of the probability tree as follows:
product of branch probabilities leading to B through U1
P1U1 ∙ B2 =
sum of all branch products leading to B
3 1
a ba b
5 3 6
= = ≈ .55
3 1 1 2 11
a ba b + a ba b
5 3 4 3
Similarly,
product of branch probabilities leading to B through U2

P1U2 ∙ B2 =
sum of all branch products leading to B
1 2
a ba b
4 3 5
= = ≈ .45
3 1 1 2 11
a ba b + a ba b
5 3 4 3
Note: We could have obtained P1U2 ∙ B2 by subtracting P1U1 ∙ B2 from 1. Why?
Matched Problem 1 Repeat Example 1, but find P1U1 ∙ W2 and P1U2 ∙ W2.

Study the probability tree below:
c M
U c1d51
a
d N
Start a1b51 a, b, c, d, e, f ? 0
e M
b e1f51
V
f N
(A) Discuss the difference between P1M ∙ U2 and P1U ∙ M2, and between P1N∙ V2
and P1V∙ N2, in terms of a, b, c, d, e, and f.
(B) Show that ac + ad + be + bf = 1. What is the significance of this result?
S In generalizing the results in Example 1, it is helpful to look at its structure in

terms of the Venn diagram shown in Figure 2. We note that U1 and U2 are mutually
B
exclusive (disjoint), and their union forms S. The following two equations can now be
U1 > B U2 > B interpreted in terms of this diagram:
U1 U2
P1U1 ¨ B2 P1U1 ¨ B2
P1U1 ∙ B2 = =
Figure 2 P1B2 P1U1 ¨ B2 + P1U2 ¨ B2
P1U2 ¨ B2 P1U2 ¨ B2
P1U2 ∙ B2 = =
P1B2 P1U1 ¨ B2 + P1U2 ¨ B2
Look over the equations and the diagram carefully.
Of course, there is no reason to stop here. Suppose that U1, U2, and U3 are three
mutually exclusive events whose union is the whole sample space S. Then, for an
S arbitrary event E in S, with P1E2 ∙ 0, the corresponding Venn diagram looks like
Figure 3, and
E
U1 > E U2 > E U3 > E P1U1 ¨ E2 P1U1 ¨ E2
P1U1 ∙ E2 = =
P1E2 P1U1 ¨ E2 + P1U2 ¨ E2 + P1U3 ¨ E2
U1 U2 U3
Similar results hold for U2 and U3.
Figure 3 Using the same reasoning, we arrive at the following famous theorem, which was
first stated by Thomas Bayes (1702–1763):
THEOREM 1 Bayes’ Formula

Let U1, U2, c, Un be n mutually exclusive events whose union is the sample
space S. Let E be an arbitrary event in S such that P1E2 ∙ 0. Then,
P1U1 ¨ E2 P1U1 ¨ E2
P1U1 ∙ E2 = =
P1E2 P1U1 ¨ E2 + P1U2 ¨ E2 + g + P1Un ¨ E2
P1E ∙ U1 2P1U1 2
=
P1E∙ U1 2P1U1 2 + P1E ∙ U2 2P1U2 2 + g + P1E∙ Un 2P1Un 2
Similar results hold for U2, U3, c, Un.
You do not need to memorize Bayes’ formula. In practice, it is often

easier to draw a probability tree and use the following:
product of branch probabilities leading to E through U1

P1U1 ∙ E2 =
sum of all branch products leading to E
Similar results hold for U2, U3, c, Un.
EXAMPLE 2 Tuberculosis Screening A new, inexpensive skin test is devised for detecting
tuberculosis. To evaluate the test before it is used, a medical researcher randomly
selects 1,000 people. Using precise but more expensive methods, it is found that
8% of the 1,000 people tested have tuberculosis. Now each of the 1,000 subjects is
given the new skin test, and the following results are recorded: The test indicates
tuberculosis in 96% of those who have it and in 2% of those who do not. Based on
these results, what is the probability of a randomly chosen person having tuberculo-
sis given that the skin test indicates the disease? What is the probability of a person
not having tuberculosis given that the skin test indicates the disease? (That is, what
is the probability of the skin test giving a false positive result ?)
SOLUTION To start, we form a tree diagram and place appropriate probabilities on
each branch:
.96 S
T T 5 tuberculosis
.08
.04 S9
T9 5 no tuberculosis
Start
S 5 skin test indicates tuberculosis
.02 S
.92 S9 5 skin test indicates no tuberculosis
T9
.98 S9
We are interested in finding P1T∙ S2, that is, the probability of a person having tuber-
culosis given that the skin test indicates the disease. Bayes’ formula for this case is
product of branch probabilities leading to S through T

P1T∙ S2 =
sum of all branch products leading to S
Substituting appropriate values from the probability tree, we obtain

1.0821.962
P1T∙ S2 = = .81
1.0821.962 + 1.9221.022
The probability of a person not having tuberculosis given that the skin test indicates
the disease, denoted by P1T ′∙ S2, is
P1T ′∙ S2 = 1 - P1T ∙ S2 = 1 - .81 = .19 P1T ∙ S2 + P1T′∙ S2 = 1
Matched Problem 2 What is the probability that a person has tuberculosis

given that the test indicates no tuberculosis is present? (That is, what is the prob-
ability of the skin test giving a false negative result?) What is the probability that
a person does not have tuberculosis given that the test indicates no tuberculosis is
present?
CONCEPTUAL INSIGHT
From a public health standpoint, which is the more serious error in Example 2
and Matched Problem 2: a false positive result or a false negative result? A false
positive result will certainly be unsettling to the subject, who might lose sleep
thinking that she has tuberculosis. But she will be sent for further testing and will
be relieved to learn it was a false alarm; she does not have the disease. A false
negative result, on the other hand, is more serious. A person who has tuberculosis
will be unaware that he has a communicable disease and so may pose a consider-
able risk to public health.
In designing an inexpensive skin test, we would expect a tradeoff between
cost and false results. We might be willing to tolerate a moderate number of false
positives if we could keep the number of false negatives at a minimum.
EXAMPLE 3 Product Defects A company produces 1,000 refrigerators a week at three plants.
Plant A produces 350 refrigerators a week, plant B produces 250 refrigerators a
week, and plant C produces 400 refrigerators a week. Production records indicate
that 5% of the refrigerators produced at plant A will be defective, 3% of those
produced at plant B will be defective, and 7% of those produced at plant C will be
defective. All the refrigerators are shipped to a central warehouse. If a refrigerator at
the warehouse is found to be defective, what is the probability that it was produced
at plant A?
SOLUTION We begin by constructing a tree diagram:
.05 D
A
.35 .95 D9 A 5 produced at plant A
.03 D B 5 produced at plant B
.25
Start B C 5 produced at plant C
.97 D9 D 5 defective
.40 .07 D D9 5 not defective
C
.93 D9
The probability that a defective refrigerator was produced at plant A is P1A∙ D2.
Bayes’ formula for this case is
product of branch probabilities leading to D through A

P1A∙ D2 =
sum of all branch products leading to D
Using the values from the probability tree, we have
1.3521.052
P1A∙ D2 =
1.3521.052 + 1.2521.032 + 1.4021.072
≈ .33
Matched Problem 3 In Example 3, what is the probability that a defective

refrigerator in the warehouse was produced at plant B? At plant C?
Exercises 7.4
Skills Warm-up Exercises S
U1 U2
W In Problems 1–6, write each expression as a quotient of integers, R
reduced to lowest terms. (If necessary, review Section B.1.)
25 35
1 2
15 25
3 7
1. 2.
1 1 1 2
+ +
3 2 4 7 13. P1U1 ∙ R2 14. P1U2 ∙ R2
1 1 1 2 1 2
3. , + 4. , + 15. P1U1 ∙ R′2 16. P1U2 ∙ R′2
3 3 2 7 4 7
4#3 2#3 B Find the probabilities in Problems 17–22 by referring to the fol-
5 4 7 4 lowing tree diagram and using Bayes’ formula. Round answers to
5. 6. two decimal places.
1#1
+ #
2#3
+ #
4 3 5 1
5 3 5 4 7 4 7 2 .7 X
K
A Find the probabilities in Problems 7–12 by referring to the tree .2 .3 Y
diagram below. .2 X
.3
Start L
.7 A .8 Y
M .6 X
.6 .5
.3 B M
Start .4 Y
.2 A
.4
N 17. P1K ∙ X2 18. P1K ∙ Y2
.8 B
19. P1L ∙ X2 20. P1L ∙ Y2
7. P1M ¨ A2 = P1M2P1A ∙ M2
21. P1M ∙ X2 22. P1M ∙ Y2
8. P1N ¨ B2 = P1N2P1B ∙ N2
Find the probabilities in Problems 23–28 by referring to the tree
9. P1A2 = P1M ¨ A2 + P1N ¨ A2 diagram below.
1
10. P1B2 = P1M ¨ B2 + P1N ¨ B2 2
5 C
P1M ¨ A2 2
1 A
3
11. P1M ∙ A2 = 4
2 D
P1M ¨ A2 + P1N ¨ A2 Start
5
2
2
5 C
P1N ¨ B2 2
12. P1N ∙ B2 = 2
3 B
P1N ¨ B2 + P1M ¨ B2 3
2
5 D
Find the probabilities in Problems 13–16 by referring to the fol- 23. P1A2 24. P1B2
lowing Venn diagram and using Bayes’ formula (assume that the
simple events in S are equally likely): 25. P1C2 26. P1D2
27. P1A ∙ C2 28. P1B ∙ D2
In Problems 29 and 30, use the probabilities in the first tree diagram In Problems 37 and 38, urn 1 contains 7 red and 3 white balls.
to find the probability of each branch of the second tree diagram. Urn 2 contains 4 red and 5 white balls. A ball is drawn from urn 1
and placed in urn 2. Then a ball is drawn from urn 2.
29. 2
1
5 B
37. If the ball drawn from urn 2 is red, what is the probability
1 A
2
4 that the ball drawn from urn 1 was red?
2
4 B9
Start
5 38. If the ball drawn from urn 2 is white, what is the probability
3
2
5 B that the ball drawn from urn 1 was white?
3
2
4 A9
2 B9
In Problems 39 and 40, refer to the following probability tree:
2
5
A c M
B U c1d51
a
A9 d N
Start Start a1b51 a, b, c, d, e, f ? 0
A e M
b e1f51
B9 V
A9 f N
30. 1
2 39. Suppose that c = e. Discuss the dependence or independence
8 D
of events U and M.
A
1
2
3
7
2 D9 40. Suppose that c = d = e = f . Discuss the dependence or
8
3 independence of events M and N.
2
1
2 8 D
3
Start B In Problems 41 and 42, two balls are drawn in succession from an
5
2 D9 urn containing m blue balls and n white balls 1m Ú 2 and n Ú 22.
8
1 1
Discuss the validity of each statement. If the statement is always
2 2
3 4 D true, explain why. If not, give a counterexample.
C
3 D9
41. (A) If the two balls are drawn with replacement, then
2
4 P1B1 ∙ B2 2 = P1B2 ∙ B1 2.
A (B) If the two balls are drawn without replacement, then
D B P1B1 ∙ B2 2 = P1B2 ∙ B1 2.
C 42. (A) If the two balls are drawn with replacement, then
Start P1B1 ∙ W2 2 = P1W2 ∙ B1 2.
A
D9 B (B) If the two balls are drawn without replacement, then
C P1B1 ∙ W2 2 = P1W2 ∙ B1 2.
C 43. If 2 cards are drawn in succession from a standard 52-card
In Problems 31–34, one of two urns is chosen at random, with one deck without replacement and the second card is a heart,
as likely to be chosen as the other. Then a ball is withdrawn from what is the probability that the first card is a heart?
the chosen urn. Urn 1 contains 1 white and 4 red balls, and urn 2
has 3 white and 2 red balls. 44. A box contains 10 balls numbered 1 through 10. Two balls are
drawn in succession without replacement. If the second ball
31. If a white ball is drawn, what is the probability that it came drawn has the number 4 on it, what is the probability that the
from urn 1? first ball had a smaller number on it? An even number on it?
32. If a white ball is drawn, what is the probability that it came
In Problems 45–50, a player is dealt two cards from a 52-card
from urn 2?
deck. If the first card is black, the player returns it to the deck
33. If a red ball is drawn, what is the probability that it came before drawing the second card. If the first card is red, the player
from urn 2? sets it aside and then draws the second card.
34. If a red ball is drawn, what is the probability that it came 45. What is the probability of drawing a red card on the second
from urn 1? draw?
46. What is the probability of drawing a black card on the second
In Problems 35 and 36, an urn contains 4 red and 5 white balls.
draw?
Two balls are drawn in succession without replacement.
47. If the second card drawn is red, what is the probability that
35. If the second ball is white, what is the probability that the the first card drawn was red?
first ball was white?
48. If the second card drawn is black, what is the probability that
36. If the second ball is red, what is the probability that the first the first card drawn was red?
ball was white?
49. If the second card drawn is red, what is the probability that
the first card drawn was black?
50. If the second card drawn is black, what is the probability that what is the probability of a randomly chosen person having
the first card drawn was black? cancer given that the test indicates cancer? Of a person having
cancer given that the test does not indicate cancer?
51. Show that P1U1 ∙ R2 + P1U1 ′∙ R2 = 1.
58. Pregnancy testing. In a random sample of 200 women who
52. If U1 and U2 are two mutually exclusive events whose union
suspect that they are pregnant, 100 turn out to be pregnant. A
is the equally likely sample space S and if E is an arbitrary
new pregnancy test given to these women indicated preg-
event in S such that P1E2 ∙ 0, show that
nancy in 92 of the 100 pregnant women and in 12 of the 100
n1U1 ¨ E2 nonpregnant women. If a woman suspects she is pregnant and
P1U1 ∙ E2 = this test indicates that she is pregnant, what is the probability
n1U1 ¨ E2 + n1U2 ¨ E2
that she is pregnant? If the test indicates that she is not preg-
nant, what is the probability that she is not pregnant?
59. Medical research. In a random sample of 1,000 people, it is
Applications found that 7% have a liver ailment. Of those who have a liver
ailment, 40% are heavy drinkers, 50% are moderate drinkers,
In the following applications, the word “probability” is often and 10% are nondrinkers. Of those who do not have a liver ail-
understood to mean “approximate empirical probability.” ment, 10% are heavy drinkers, 70% are moderate drinkers, and
53. Employee screening. The management of a company finds 20% are nondrinkers. If a person is chosen at random and he
that 30% of the administrative assistants hired are unsatisfac- or she is a heavy drinker, what is the probability of that person
tory. The personnel director is instructed to devise a test that having a liver ailment? What is the probability for a nondrinker?
will improve the situation. One hundred employed admin- 60. Tuberculosis screening. A test for tuberculosis was given to
istrative assistants are chosen at random and are given the 1,000 subjects, 8% of whom were known to have tuberculo-
newly constructed test. Out of these, 90% of the satisfactory sis. For the subjects who had tuberculosis, the test indicated
administrative assistants pass the test and 20% of the unsatis- tuberculosis in 95% of the subjects, was inconclusive for 3%,
factory administrative assistants pass. Based on these results, and indicated no tuberculosis in 2%. For the subjects who did
if a person applies for a job, takes the test, and passes it, what not have tuberculosis, the test indicated tuberculosis in 4%
is the probability that he or she is a satisfactory administrative of the subjects, was inconclusive for 6%, and indicated no
assistant? If the applicant fails the test, what is the probability tuberculosis in the remaining 90%. What is the probability of
that he or she is a satisfactory administrative assistant? a randomly selected person having tuberculosis given that the
54. Employee rating. A company has rated 75% of its em- test indicates tuberculosis? Of not having tuberculosis given
ployees as satisfactory and 25% as unsatisfactory. Personnel that the test was inconclusive?
records indicate that 80% of the satisfactory workers had pre- 61. Police science. A new lie-detector test has been devised
vious work experience, while only 40% of the unsatisfactory and must be tested before it is used. One hundred people are
workers had any previous work experience. If a person with selected at random, and each person draws a card from a box
previous work experience is hired, what is the probability that of 100 cards. Half the cards instruct the person to lie, and the
this person will be a satisfactory employee? If a person with others instruct the person to tell the truth. Of those who lied,
no previous work experience is hired, what is the probability 80% fail the new lie-detector test (that is, the test indicates
that this person will be a satisfactory employee? lying). Of those who told the truth, 5% failed the test. What is
55. Product defects. A manufacturer obtains GPS systems from the probability that a randomly chosen subject will have lied
three different subcontractors: 20% from A, 40% from B, and given that the subject failed the test? That the subject will not
40% from C. The defective rates for these subcontractors are have lied given that the subject failed the test?
1%, 3%, and 2%, respectively. If a defective GPS system is 62. Politics. In a given region, records show that of the registered
returned by a customer, what is the probability that it came voters, 42% are members of the Conservative party, 39% are
from subcontractor A? From B? From C? members of the Labor party, and 19% are members of other
56. Product defects. A store sells three types of flash drives: political parties. In the EU Referendum, 45% of Conserva-
brand A, brand B, and brand C. Of the flash drives it sells, tive voters, 65% of Labor voters, and 50% of the others voted
60% are brand A, 25% are brand B, and 15% are brand C. in favor of remaining in the European Union. If a registered
The store has found that 20% of the brand A flash drives, voter chosen at random is found to have voted in favor of
15% of the brand B flash drives, and 5% of the brand C flash remaining in the European Union, what is the probability that
drives are returned as defective. If a flash drive is returned the voter is a member of the Conservative party? Member of
as defective, what is the probability that it is a brand A flash the Labor party? Member of another political party?
drive? A brand B flash drive? A brand C flash drive?
57. Cancer screening. A new, simple test has been developed to Answers to Matched Problems
detect a particular type of cancer. The test must be evaluated
4 15
before it is used. A medical researcher selects a random sample 1. P1U1 ∙ W2 = ≈ .21; P1U2 ∙ W2 = ≈ .79
of 1,000 adults and finds (by other means) that 2% have this 19 19
type of cancer. Each of the 1,000 adults is given the test, and 2. P1T ∙ S′2 = .004; P1T ′∙ S ′2 = .996
it is found that the test indicates cancer in 98% of those who 3. P1B ∙ D2 ≈ .14; P1C∙ D2 ≈ .53
have it and in 1% of those who do not. Based on these results,
SECTION 7.5 Random Variable, Probability Distribution, and Expected Value 421
7.5 Random Variable, Probability Distribution, and Expected Value

■■ Random Variable and Probability Random Variable and Probability Distribution
Distribution
When performing a random experiment, a sample space S is selected in such a way
■■ Expected Value of a Random that all probability problems of interest relative to the experiment can be solved. In
Variable many situations we may not be interested in each simple event in the sample space
■■ Decision Making and Expected S but in some numerical value associated with the event. For example, if 3 coins are
Value tossed, we may be interested in the number of heads that turn up rather than in the
particular pattern that turns up. Or, in selecting a random sample of students, we may
be interested in the proportion that are women rather than which particular students
are women. In the same way, a “craps” player is usually interested in the sum of the
dots on the showing faces of the dice rather than the pattern of dots on each face.
In each of these examples, there is a rule that assigns to each simple event in S a single
real number. Mathematically speaking, we are dealing with a function (see Section 1.1).
Historically, this particular type of function has been called a “random variable.”
DEFINITION Random Variable

A random variable is a function that assigns a numerical value to each simple
event in a sample space S.
The term random variable is an unfortunate choice, since it is neither random

nor a variable—it is a function with a numerical value, and it is defined on a sample
Table 1 Number of Heads in the space. But the terminology has stuck and is now standard. Capital letters, such as X,
Toss of 3 Coins are used to represent random variables.
Let us return to the experiment of tossing 3 coins. A sample space S of equally
Sample Space Number of
S Heads X1ei 2 likely simple events is indicated in Table 1. Suppose that we are interested in the
e1: TTT 0
number of heads 10, 1, 2, or 32 appearing on each toss of the 3 coins and the prob-
e2: TTH 1 ability of each of these events. We introduce a random variable X (a function) that
e3: THT 1 indicates the number of heads for each simple event in S (see the second column in
e4: HTT 1 Table 1). For example, X1e1 2 = 0, X1e2 2 = 1, and so on. The random variable X as-
e5: THH 2 signs a numerical value to each simple event in the sample space S.
e6: HTH 2 We are interested in the probability of the occurrence of each image or range
e7: HHT 2 value of X, that is, in the probability of the occurrence of 0 heads, 1 head, 2 heads, or
e8: HHH 3 3 heads in the single toss of 3 coins. We indicate this probability by
p1x2 where x ∊ 50, 1, 2, 36
The function p is called the probability distribution* of the random variable X.
What is p122, the probability of getting exactly 2 heads on the single toss of 3
coins? “Exactly 2 heads occur” is the event
E = 5THH, HTH, HHT6
Thus,
n1E2 3
p122 = =
Table 2 Probability Distribution
n1S2 8
Number of Proceeding similarly for p102, p112, and p132, we obtain the probability distribu-
0 1 2 3
Heads x tion of the random variable X presented in Table 2. Probability distributions are also
Probability 1 3 3 1 represented graphically, as shown in Figure 1. The graph of a probability distribution
p1x2 8 8 8 8 is often called a histogram.
*The probability distribution p of the random variable X is defined by p1x2 = P1 5ei ∊ S∙ X1ei 2 = x6 2,
which, because of its cumbersome nature, is usually simplified to p1x2 = P1X = x2 or simply p1x2. We
will use the simplified notation.
p(x) 5 P(X 5 x)
3
2
8
2
2
8
1
2
8
0 x
0 1 2 3
Number of heads
Figure 1 Histogram for a probability distribution
Note from Table 2 or Figure 1 that

1. 0 … p1x2 … 1, x ∊ 50, 1, 2, 36
3 3
2. p102 + p112 + p122 + p132 = 18 + 8 + 8 + 18 = 1
These are general properties that any probability distribution of a random variable X
associated with a finite sample space must have.
THEOREM 1 Probability Distribution of a Random Variable X

The probability distribution of a random variable X, denoted by P1X = x2 ∙
p1x2, satisfies
1. 0 … p1x2 … 1, x ∊ 5x1, x2, c, xn 6
2. p1x1 2 + p1x2 2 + g + p1xn 2 = 1
where 5x1, x2, c, xn 6 are the (range) values of X (see Fig. 2).
Figure 2 illustrates the process of forming a probability distribution of a random

variable.
Sample space 5 Range of random variable 5 Range of probability

Domain of random variable Domain of probability distribution distribution
S X p
Random Random
e1 e2 e3 e4 variable x1 x2 variable p1 p2
e5 e6 e7 x3 x4 x5 p3 p4 p5
Outcomes of an Probabilities assigned to

experiment—not Numerical values assigned to
values of the random
necessarily numerical outcomes in S
variable
values
Figure 2 Probability distribution of a random variable for a finite sample space
Expected Value of a Random Variable

Suppose that the experiment of tossing 3 coins was repeated many times. What would
be the average number of heads per toss (the total number of heads in all tosses di-
vided by the total number of tosses)? Consulting the probability distribution in Table 2
or Figure 1, we would expect to toss 0 heads 18 of the time, 1 head 38 of the time,
2 heads 38 of the time, and 3 heads 18 of the time. In the long run, we would expect the
average number of heads per toss of the 3 coins, or the expected value E1X2, to be
given by
1 3 3 1 12
E1X2 = 0 a b + 1 a b + 2 a b + 3 a b = = 1.5
8 8 8 8 8
It is important to note that the expected value is not a value that will necessarily
occur in a single experiment (1.5 heads cannot occur in the toss of 3 coins), but it is
an average of what occurs over a large number of experiments. Sometimes we will
toss more than 1.5 heads and sometimes less, but if the experiment is repeated many
times, the average number of heads per experiment should be close to 1.5.
We now make the preceding discussion more precise through the following defi-
nition of expected value:
DEFINITION Expected Value of a Random Variable X

Given the probability distribution for the random variable X,
xi x1 x2 c xn
pi p1 p2 c pn
where pi = p1xi 2, we define the expected value of X, denoted E1 X2 , by the for-

mula
E1X2 = x1p1 + x2p2 + g + xnpn
We again emphasize that the expected value is not the outcome of a single
experiment, but a long-run average of outcomes of repeated experiments. The ex-
pected value is the weighted average of the possible outcomes, each weighted by its
probability.
PROCEDURE Steps for Computing the Expected Value of a Random Variable X
Step 1 Form the probability distribution of the random variable X.

Step 2 Multiply each image value of X, xi, by its corresponding probability of oc-
currence pi; then add the results.
EXAMPLE 1 Expected Value What is the expected value (long-run average) of the number of
dots facing up for the roll of a single die?
SOLUTION If we choose
S = 51, 2, 3, 4, 5, 66
as our sample space, then each simple event is a numerical outcome reflecting our
interest, and each is equally likely. The random variable X in this case is just the
identity function (each number is associated with itself). The probability distribu-
tion for X is
xi 1 2 3 4 5 6
1 1 1 1 1 1
pi
6 6 6 6 6 6
Therefore,
1 1 1 1 1 1
E1X2 = 1 a b + 2 a b + 3 a b + 4 a b + 5 a b + 6 a b
6 6 6 6 6 6
21
= = 3.5
6
Matched Problem 1 Suppose that the die in Example 1 is not fair and we
obtain (empirically) the following probability distribution for X:
xi 1 2 3 4 5 6
3Note: Sum = 1.4
pi .14 .13 .18 .20 .11 .24
What is the expected value of X?

From Example 1 we can conclude that the probability is 0 that a single roll of a fair
die will equal the expected value for a roll of a die (the number of dots facing up is
never 3.5). What is the probability that the sum for a single roll of a pair of dice will
equal the expected value of the sum for a roll of a pair of dice?
EXAMPLE 2 Expected Value A carton of 20 laptop batteries contains 2 defective ones. A ran-
dom sample of 3 is selected from the 20 and tested. Let X be the random variable
associated with the number of defective batteries found in a sample.
(A) Find the probability distribution of X.
(B) Find the expected number of defective batteries in a sample.
SOLUTION
(A) The number of ways of selecting a sample of 3 from 20 (order is not important)
is 20C3. This is the number of simple events in the experiment, each as likely
as the other. A sample will have either 0, 1, or 2 defective batteries. These are
the values of the random variable in which we are interested. The probability
distribution is computed as follows:
18C3 2C1 18C2

# #
2C2 18C1
p102 = ≈ .716 p112 = ≈ .268 p122 = ≈ .016
20C3 20C3 20C3
We summarize these results in a table:

xi 0 1 2
pi .716 .268 .016
3Note: .716 + .268 + .016 = 1.4
(B) The expected number of defective batteries in a sample is readily computed as

follows:
E1X2 = 1021.7162 + 1121.2682 + 1221.0162 = .3
The expected value is not one of the random variable values; rather, it is a num-
ber that the average number of defective batteries in a sample would approach
as the experiment is repeated without end.
Matched Problem 2 Repeat Example 2 using a random sample of 4.

EXAMPLE 3 Expected Value of a Game A spinner device is numbered from 0 to 5, and each
of the 6 numbers is as likely to come up as any other. A player who bets $1 on any
given number wins $4 (and gets the $1 bet back) if the pointer comes to rest on the
chosen number; otherwise, the $1 bet is lost. What is the expected value of the game
(long-run average gain or loss per game)?
SOLUTION The sample space of equally likely events is
S = 50, 1, 2, 3, 4, 56
Each sample point occurs with a probability of 16. The random variable X assigns $4
to the winning number and - $1 to each of the remaining numbers. So the probabil-
ity of winning $4 is 16 and of losing $1 is 56. We form the probability distribution for
X, called a payoff table, and compute the expected value of the game:
Payoff Table (Probability Distribution for X)

xi $4 - $1
1 5
pi
6 6
1 5 1
E1X2 = $4 a b + 1 - $12 a b = - $ ≈ - $0.1667 ≈ - 17¢ per game
6 6 6
In the long run, the player will lose an average of about 17¢ per game.
Matched Problem 3 Repeat Example 3 with the player winning $5 instead of

$4 if the chosen number turns up. The loss is still $1 if any other number turns up.
Is this a fair game?
The game in Example 3 is not fair: The player tends to lose money in the long
run. A game is fair if the expected value E1X2 is equal to 0; that is, the player neither
wins nor loses money in the long run. The fair games discussed in Section 7.2 are fair
according to this definition, because their payoff tables have the following form:
Payoff Table
xi $b - $a
a b
pi a + b a + b
a b
So E1X2 = b a b + 1 - a2 = 0
a + b a + b
EXAMPLE 4 Expected Value and Insurance Suppose you are interested in insuring a car video
system for $2,000 against theft. An insurance company charges a premium of $225
for coverage for 1 year, claiming an empirically determined probability of .1 that the
system will be stolen sometime during the year. What is your expected return from
the insurance company if you take out this insurance?
SOLUTION This is actually a game of chance in which your stake is $225. You have
a .1 chance of receiving $1,775 from the insurance company ($2,000 minus your
stake of $225) and a .9 chance of losing your stake of $225. What is the expected
value of this “game”? We form a payoff table (the probability distribution for X) and
compute the expected value:
Payoff Table
xi $1,775 - $225
pi .1 .9
E1X2 = 1$1,77521.12 + 1 - $22521.92 = - $25

This means that if you insure with this company over many years and circumstances
remain the same, you would have an average net loss to the insurance company of
$25 per year.
Matched Problem 4 Find the expected value in Example 4 from the insurance
company’s point of view.
CONCEPTUAL INSIGHT
Suppose that in a class of 10 students, the scores on the first exam are 85, 73, 82,
65, 95, 85, 73, 75, 85, and 75. To compute the class average (mean), we add the
scores and divide by the number of scores:
85 + 73 + 82 + 65 + 95 + 85 + 73 + 75 + 85 + 75 793
= = 79.3
10 10
Because 1 student scored 95, 3 scored 85, 1 scored 82, 2 scored 75, 2 scored 73,
and 1 scored 65, the probability distribution of an exam score, for a student cho-
sen at random from the class, is as follows:
xi 65 73 75 82 85 95
pi .1 .2 .2 .1 .3 .1
The expected value of the probability distribution is

1 2 2 1 3 1 793
65 a b + 73 a b + 75 a b + 82 a b + 85 a b + 95 a b = = 79.3
10 10 10 10 10 10 10
By comparing the two computations, we see that the mean of a data set is just the
expected value of the corresponding probability distribution.
Decision Making and Expected Value

We conclude this section with an example in decision making.
EXAMPLE 5 Decision Analysis An outdoor concert featuring a popular musical group is sched-
uled for a Sunday afternoon in a large open stadium. The promoter, worrying about
being rained out, contacts a long-range weather forecaster who predicts the chance
of rain on that Sunday to be .24. If it does not rain, the promoter is certain to net
$100,000; if it does rain, the promoter estimates that the net will be only $10,000.
An insurance company agrees to insure the concert for $100,000 against rain at a
premium of $20,000. Should the promoter buy the insurance?
SOLUTION The promoter has a choice between two courses of action; A1: Insure
and A2: Do not insure. As an aid in making a decision, the expected value is com-
puted for each course of action. Probability distributions are indicated in the payoff
table (read vertically):
Payoff Table
A1: Insure A2: Do Not Insure
pi xi xi
.24 (rain) $90,000 $10,000
.76 (no rain) $80,000 $100,000
Note that the $90,000 entry comes from the insurance company’s payoff 1$100,0002
minus the premium 1$20,0002 plus gate receipts 1$10,0002. The reasons for the
other entries should be obvious. The expected value for each course of action is
computed as follows:
A1: Insure A2: Do Not Insure

E1X2 = x1 p1 + x2 p2 E1X2 = 1 $10,0002 1.242 + 1 $100,0002 1.762
= 1 $90,0002 1.242 + 1 $80,0002 1.762 = $78,400
= $82,400
It appears that the promoter’s best course of action is to buy the insurance at $20,000.
The promoter is using a long-run average to make a decision about a single event—
a common practice in making decisions in areas of uncertainty.
Matched Problem 5 In Example 5, what is the insurance company’s expected

value if it writes the policy?
Exercises 7.5
Skills Warm-up Exercises 10. You draw and keep a single bill from a hat that contains a $1,
$5, $10, $20, $50, and $100 bill. What is the expected value
W In Problems 1–8, if necessary, review Section A.1. of the game to you?
1. Find the average (mean) of the exam scores 73, 89, 45, 82, 11. You draw and keep a single coin from a bowl that contains
and 66. 15 pennies, 10 dimes, and 25 quarters. What is the expected
2. Find the average (mean) of the exam scores 78, 64, 97, 60, value of the game to you?
86, and 83. 12. You draw and keep a single coin from a bowl that contains
3. Find the average (mean) of the exam scores in Problem 1, if 120 nickels and 80 quarters. What is the expected value of
4 points are added to each score. the game to you?
4. Find the average (mean) of the exam scores in Problem 2, if 13. You draw a single card from a standard 52-card deck. If it
6 points are subtracted from each score. is red, you win $50. Otherwise you get nothing. What is the
expected value of the game to you?
5. Find the average (mean) of the exam scores in Problem 1, if
each score is multiplied by 2. 14. You draw a single card from a standard 52-card deck. If it is a
face card, you win $130. Otherwise you get nothing. What is
6. Find the average (mean) of the exam scores in Problem 2, if the expected value of the game to you?
each score is divided by 2.
15. In tossing 2 fair coins, what is the expected number of heads?
A 7. If the probability distribution for the random variable X is
given in the table, what is the expected value of X? 16. In a family with 2 children, excluding multiple births and
assuming that a boy is as likely as a girl at each birth, what is
xi -3 0 4
the expected number of boys?
pi .3 .5 .2
17. A fair coin is flipped. If a head turns up, you win $1. If a
8. If the probability distribution for the random variable X is
tail turns up, you lose $1. What is the expected value of the
given in the table, what is the expected value of X?
game? Is the game fair?
xi -2 -1 0 1 2
18. Repeat Problem 17, assuming an unfair coin with the prob-
pi .1 .2 .4 .2 .1 ability of a head being .4 and a tail being .6.
9. You draw and keep a single bill from a hat that contains a
B 19. After paying $4 to play, a single fair die is rolled, and you are
$5, $20, $50, and $100 bill. What is the expected value
paid back the number of dollars corresponding to the number
of the game to you?
of dots facing up. For example, if a 5 turns up, $5 is returned A1 A2

to you for a net gain, or payoff, of $1; if a 1 turns up, $1 is re- pi xi xi
turned for a net gain of - $3; and so on. What is the expected
.1 - $200 - $100
value of the game? Is the game fair?
.2 $100 $200
20. Repeat Problem 19 with the same game costing $3.50 for
.4 $400 $300
each play.
.3 $100 $200
21. Two coins are flipped. You win $5 if either 2 heads or 2 tails
turn up; you lose $10 if a head and a tail turn up. What is the 34. The payoff table for three possible courses of action is given
expected value of the game? below. Which of the three actions will produce the largest
22. In Problem 21, for the game to be fair, how much should you expected value? What is it?
lose if a head and a tail turn up?
A1 A2 A3
23. A friend offers the following game: She wins $1 from you if, Pi xi xi xi
on four rolls of a single die, a 6 turns up at least once; other-
.2 $ 500 $ 400 $ 300
wise, you win $1 from her. What is the expected value of the
game to you? To her? .4 $1,200 $1,100 $1,000
.3 $1,200 $1,800 $1,700
24. On three rolls of a single die, you will lose $10 if a 5 turns
up at least once, and you will win $7 otherwise. What is the .1 $1,200 $1,800 $2,400
expected value of the game?
35. Roulette wheels in Nevada generally have 38 equally spaced
25. A single die is rolled once. You win $5 if a 1 or 2 turns up and slots numbered 00, 0, 1, 2, c, 36. A player who bets $1 on
$10 if a 3, 4, or 5 turns up. How much should you lose if a any given number wins $35 (and gets the bet back) if the ball
6 turns up in order for the game to be fair? Describe the comes to rest on the chosen number; otherwise, the $1 bet is
steps you took to arrive at your answer. lost. What is the expected value of this game?
26. A single die is rolled once. You lose $12 if a number divis- 36. In roulette (see Problem 35), the numbers from 1 to 36 are
ible by 3 turns up. How much should you win if a number evenly divided between red and black. A player who bets $1
not divisible by 3 turns up in order for the game to be fair? on black wins $1 (and gets the $1 bet back) if the ball comes
Describe the process and reasoning used to arrive at your to rest on black; otherwise (if the ball lands on red, 0, or 00),
answer. the $1 bet is lost. What is the expected value of the game?
27. A pair of dice is rolled once. Suppose you lose $10 if a 37. A game has an expected value to you of $100. It costs $100
7 turns up and win $11 if an 11 or 12 turns up. How much to play, but if you win, you receive $100,000 (including your
should you win or lose if any other number turns up in order $100 bet) for a net gain of $99,900. What is the probability of
for the game to be fair? winning? Would you play this game? Discuss the factors that
28. A coin is tossed three times. Suppose you lose $3 if 3 heads would influence your decision.
appear, lose $2 if 2 heads appear, and win $3 if 0 heads ap- 38. A game has an expected value to you of - $1.50. It costs $5
pear. How much should you win or lose if 1 head appears in to play, but if you win, you receive $1,000 (including your $5
order for the game to be fair? bet) for a net gain of $995. What is the probability of winning?
29. A card is drawn from a standard 52-card deck. If the card Would you play this game? Discuss the factors that would
is a king, you win $10; otherwise, you lose $1. What is the influence your decision.
expected value of the game? C 39. Five thousand tickets are sold at $1 each for a charity raffle.
30. A card is drawn from a standard 52-card deck. If the card is Tickets will be drawn at random and monetary prizes award-
a diamond, you win $10; otherwise, you lose $4. What is the ed as follows: 1 prize of $500; 3 prizes of $100, 5 prizes of
expected value of the game? $20, and 20 prizes of $5. What is the expected value of this
raffle if you buy 1 ticket?
31. A 5-card hand is dealt from a standard 52-card deck. If the
hand contains at least one king, you win $10; otherwise, you 40. Two thousand raffle tickets are sold at $5 each for a local com-
lose $1. What is the expected value of the game? munity center benefit. Prizes are awarded as follows: 1 prize of
$2,500, 3 prizes of $1,000, and 20 prizes of $200. What is the
32. A 5-card hand is dealt from a standard 52-card deck. expected value of this raffle if you purchase 1 ticket?
If the hand contains at least one diamond, you win $10;
otherwise, you lose $4. What is the expected value of the 41. A box of 10 flashbulbs contains 3 defective bulbs. A random
game? sample of 2 is selected and tested. Let X be the random variable
associated with the number of defective bulbs in the sample.
33. The payoff table for two courses of action, A1 or A2, is given
below. Which of the two actions will produce the largest (A) Find the probability distribution of X.
expected value? What is it? (B) Find the expected number of defective bulbs in a sample.
42. A box of 8 flashbulbs contains 3 defective bulbs. A random 48. A 3-card hand is dealt from a standard deck. You win $100
sample of 2 is selected and tested. Let X be the random for each king in the hand. If the game is fair, how much
variable associated with the number of defective bulbs in a should you lose if the hand contains no kings?
sample.
(A) Find the probability distribution of X.
(B) Find the expected number of defective bulbs in a sample.
43. One thousand raffle tickets are sold at $1 each. Three tickets
Applications
will be drawn at random (without replacement), and each will 49. Insurance. The annual premium for a $10,000 insurance
pay $200. Suppose you buy 5 tickets. policy against the theft of a sculpture is $125. If the (empiri-
(A) Create a payoff table for 0, 1, 2, and 3 winning tickets cal) probability that the sculpture will be stolen during the
among the 5 tickets you purchased. (If you do not have year is .005, what is your expected return from the insurance
any winning tickets, you lose $5; if you have 1 winning company if you take out this insurance?
ticket, you net $195 since your initial $5 will not be
returned to you; and so on.) 50. Insurance. An insurance company charges an annual pre-
mium of $75 for a $200,000 insurance policy against a house
(B) What is the expected value of the raffle to you? burning down. If the (empirical) probability that a house
44. Repeat Problem 43 with the purchase of 10 tickets. burns down in a given year is .0003, what is the expected
value of the policy to the insurance company?
45. To simulate roulette on a graphing calculator, a random integer
between - 1 and 36 is selected (- 1 represents 00; see 51. Decision analysis. After careful testing and analysis, an oil
Problem 35). The command in Figure A simulates 200 games. company is considering drilling in two different sites. It is
estimated that site A will net $30 million if successful (prob-
ability .2) and lose $3 million if not (probability .8); site B
will net $70 million if successful (probability .1) and lose $4
million if not (probability .9). Which site should the company
choose according to the expected return for each site?
52. Decision analysis. Repeat Problem 51, assuming that ad-
ditional analysis caused the estimated probability of success
in field B to be changed from .1 to .11.
53. Genetics. Suppose that at each birth, having a girl is not as
(A) likely as having a boy. The probability assignments for the
number of boys in a 3-child family are approximated empiri-
cally from past records and are given in the table. What is the
12
expected number of boys in a 3-child family?
Number of Boys
xi pi
0 .12
0 38
24 1 .36
2 .38
(B)
3 .14
(A) Use the statistical plot in Figure B to determine the net
gain or loss of placing a $1 bet on the number 13 in each 54. Genetics. A pink-flowering plant is of genotype RW. If
of the 200 games. two such plants are crossed, we obtain a red plant (RR) with
probability .25, a pink plant (RW or WR) with probability
(B) Compare the results of part (A) with the expected value .50, and a white plant (WW) with probability .25, as shown
of the game. in the table. What is the expected number of W genes present
(C) Use a graphing calculator to simulate betting $1 on the in a crossing of this type?
number 7 in each of 500 games of roulette and compare
the simulated and expected gains or losses. Number of
46. Use a graphing calculator to simulate the results of placing a $1 W Genes Present
bet on black in each of 400 games of roulette (see Problems 36 xi pi
and 45) and compare the simulated and expected gains or losses. 0 .25
47. A 3-card hand is dealt from a standard deck. You win $20 1 .50
for each diamond in the hand. If the game is fair, how much 2 .25
should you lose if the hand contains no diamonds?
55. Lottery. A £2 UK Lotto ticket has a 1>10.3 probability of 2. (A) xi 0 1 2

winning one Lotto Lucky Dip (worth £2), a 1>97 probabil- pi .632 .337 .032*
ity of winning £25, a 1>2,180 probability of winning £100,
*Note: Due to roundoff error,
a 1>144,145 probability of winning £1,000, a 1>7,509,579
sum = 1.001 ≈ 1.
probability of winning £50,000, and a 1>45,057,474 prob-
ability of winning the jackpot. If the jackpot is currently (B) .4
£5,000,000, what is the expected value of an UK Lotto 3. E1X2 = $0; the game is fair
ticket?
4. E1X2 = 1 - $1,7752 1.12 + 1 $2252 1.92 = $25 (This
56. Lottery. Repeat Problem 55, assuming that the jackpot is amount, of course, is necessary to cover expenses and profit.)
currently £22,000,000.
5. E1X2 = 1 - $80,0002 1.242 + 1 $20,0002 1.762 = - $4,000
(This means that the insurance company had other information
Answers to Matched Problems regarding the weather than the promoter had; otherwise, the
1. E1X2 = 3.73 company would not have written this policy.)

7.1 Sample Spaces, Events, and Probability EXAMPLES
• Probability theory is concerned with random experiments (such as tossing a coin or rolling a pair of
dice) for which different outcomes are obtained no matter how carefully the experiment is repeated under
the same conditions.
• The set S of all outcomes of a random experiment is called a sample space. The subsets of S are called Ex. 1, p. 374
events. An event that contains only one outcome is called a simple event or simple outcome. Events
Ex. 2, p. 375
that contain more than one outcome are compound events. We say that an event E occurs if any of the
simple events in E occurs. Ex. 3, p. 376
• If S = 5e1, e2, c, en 6 is a sample space for an experiment, an acceptable probability assignment is
an assignment of real numbers P1ei 2 to simple events such that
0 … P1ei 2 … 1 and P1e1 2 + P1e2 2 + g + P1en 2 = 1
• Each number P1ei 2 is called the probability of the event ei. The probability of an arbitrary event E, Ex. 4, p. 378
denoted by P1E2, is the sum of the probabilities of the simple events in E. If E is the empty set, then
P1E2 = 0.
• Acceptable probability assignments can be made using a theoretical approach or an empirical approach. Ex. 6, p. 381
If an experiment is conducted n times and event E occurs with frequency f1E2, then the ratio f1E2 >n is
called the relative frequency of the occurrence of E in n trials, or the approximate empirical prob-
ability of E. The empirical probability of E is the number (if it exists) that f1E2 >n approaches as n gets
larger and larger.
• If the equally likely assumption is made, each simple event of the sample space S = 5e1, e2,.., en 6 is Ex. 5, p. 381
assigned the same probability, namely, 1>n. Theorem 1 (p. 380) gives the probability of arbitrary events Ex. 7, p. 382
under the equally likely assumption. Ex. 8, p. 382
7.2 Union, Intersection, and Complement of Events; Odds
• Let A and B be two events in a sample space. Then A ∪ B = 5x ∙ x ∊ A or x ∊ B6 is the union of A and Ex. 1, p. 387
B; A ¨ B = 5x ∙ x ∊ A and x ∊ B6 is the intersection of A and B.
• Events whose intersection is the empty set are said to be mutually exclusive or disjoint.
• The probability of the union of two events is given by Ex. 2, p. 388
P1A ∪ B2 = P1A2 + P1B2 - P1A ¨ B2 Ex. 3, p. 390
• The complement of event E, denoted by E′, consists of those elements of S that do not belong to E: Ex. 4, p. 391
Ex. 5, p. 391
P1E′2 = 1 - P1E2
• The language of odds is sometimes used, as an alternative to the language of probability, to describe Ex. 6, p. 393
the likelihood of an event. If P1E2 is the probability of E, then the odds for E are P1E2 >P1E′2 [usu- Ex. 7, p. 394
ally expressed as a ratio of whole numbers and read as “P1E2 to P1E′2”], and the odds against E are Ex. 8, p. 394
P1E′2 >P1E2.
• If the odds for an event E are a/b, then
a
P1E2 =
a + b
7.3 Conditional Probability, Intersection, and Independence

• If A and B are events in a sample space S, and P1B2 ∙ 0, then the conditional probability of Ex. 1, p. 401
A given B is defined by
Ex. 2, p. 401
P1A ¨ B2
P1A ∙ B2 =
P1B2
• By solving this equation for P1A ¨ B2 we obtain the product rule (Theorem 1, p. 402): Ex. 3, p. 402
Ex. 4, p. 403
P1A ¨ B2 = P1B2P1A ∙ B2 = P1A2P1B ∙ A2 Ex. 5, p. 405
Ex. 6, p. 407
• Events A and B are independent if P1A ¨ B2 = P1A2P1B2. Ex. 7, p. 408
• Theorem 2 (p. 411) gives a test for independence. Ex. 8, p. 409
7.4 Bayes’ Formula
• Let U1, U2, c, Un be n mutually exclusive events whose union is the sample space S. Let E be an arbi- Ex. 1, p. 413
trary event in S such that P1E2 ∙ 0. Then Ex. 2, p. 416
P1E ∙ U1 2P1U1 2 Ex. 3, p. 417
P1U1 ∙ E2 =
P1E ∙ U1 2P1U1 2 + P1E ∙ U2 2P1U2 2 + g + P1E ∙ Un 2P1Un 2
product of branch probabilities leading to E through U1
=
sum of all branch products leading to E
Similar results hold for U2, U3, c, Un. This formula is called Bayes’ formula.
7.5 Random Variable, Probability Distribution, and Expected Value

• A random variable X is a function that assigns a numerical value to each simple event in a sample space S.
• The probability distribution of X assigns a probability p1x2 to each range element x of X: p1x2 is the
sum of the probabilities of the simple events in S that are assigned the numerical value x.
• If a random variable X has range values x1, x2, c, xn that have probabilities p1, p2, c, pn, respectively, Ex. 1, p. 423
then the expected value of X, denoted E1X2, is defined by
Ex. 2, p. 424
E1X2 = x1p1 + x2p2 + g + xnpn Ex. 3, p. 425
Ex. 4, p. 425
• Suppose the xi’s are payoffs in a game of chance. If the game is played a large number of times, the
expected value approximates the average win per game. Ex. 5, p. 426
Review Exercises
Work through all the problems in this chapter review and check 17. Are S and X independent?
your answers in the back of the book. Answers to all problems are
Answer Problems 18–25 using the following probability tree:
there along with section numbers in italics to indicate where each
type of problem is discussed. Where weaknesses show up, review .2 B
appropriate sections in the text. A
.4
.8 B9
1. In a single deal of 5 cards from a standard 52-card deck, what Start
is the probability of being dealt 5 clubs? .3 B
.6
2. Mashed potatoes and pizza form part of a 10-item menu at a A9
restaurant. If Peter visits the restaurant for lunch and supper .7 B9
and orders randomly on both occasions, what is the prob-
18. P1A2 19. P1B ∙ A2
ability that he orders mashed potatoes for lunch and pizza for
supper if he has decided not to have the same item for both 20. P1B ∙ A′2 21. P1A ¨ B2
lunch and supper?
22. P1A′ ¨ B2 23. P1B2
3. Each of the first 10 letters of the alphabet is printed on a
24. P1A ∙ B2 25. P1A ∙ B′2
separate card. What is the probability of drawing 3 cards and
getting the code word dig by drawing d on the first draw, i on 26. (A) If 10 out of 32 students in a class were born in June, July,
the second draw, and g on the third draw? What is the prob- or August, what is the approximate empirical probability
ability of being dealt a 3-card hand containing the letters d, i, of any student being born in June, July, or August?
and g in any order? (B) If one is as likely to be born in any of the 12 months of
a year as any other, what is the theoretical probability of
4. A drug has side effects for 60 out of 1,200 people in a test.
being born in either June, July, or August?
What is the approximate empirical probability that a person
using the drug will have side effects? (C) Discuss the discrepancy between the answers to parts
(A) and (B).
5. A spinning device has 5 numbers, 1, 2, 3, 4, and 5, each
In Problems 27 and 28, a sample space S is described. Would it be
as likely to turn up as the other. A person pays $3 and then
reasonable to make the equally likely assumption? Explain.
receives back the dollar amount corresponding to the number
turning up on a single spin. What is the expected value of the 27. A 3-card hand is dealt from a standard deck. We are interest-
game? Is the game fair? ed in the number of red cards in the hand, so an appropriate
sample space is S = 50, 1, 2, 36.
6. If A and B are events in a sample space S and
28. A 3-card hand is dealt from a standard deck. We are inter-
P1A′2 = .45, P1B2 = .6 and P1A ∪ B2 = .85, find
ested in whether there are more red cards or more black cards
(A) P1A2 in the hand, so an appropriate sample space is S = 5R, B6.
(B) P1A ¨ B2 29. A player tosses two coins and receives $5 if 2 heads turn up,
7. A spinner lands on R with probability .3, on G with probabil- loses $4 if 1 head turns up, and wins $2 if 0 heads turn up.
ity .5, and on B with probability .2. Find the probability and Compute the expected value of the game. Is the game fair?
odds for the spinner landing on either R or G.
30. A spinning device has 3 numbers, 2, 5, and 9, each as likely
8. If in repeated rolls of two fair dice the odds for rolling a sum to turn up as the other. If the device is spun twice, what is the
of 8 before rolling a sum of 7 are 5 to 6, then what is the probability that
probability of rolling a sum of 8 before rolling a sum of 7? (A) Two prime numbers are generated?
Answer Problems 9–17 using the table of probabilities shown below. (B) A larger number is generated first, followed by
X Y Z Totals a smaller one?
S .10 .25 .15 .50 31. In a single draw from a standard 52-card deck, what are the
probability and odds for drawing
T .05 .20 .02 .27
R .05 .15 .03 .23
(A) A jack or a queen?
Totals .20 .60 .20 1.00
(B) A jack or a spade?
(C) A card other than an ace?
9. Find P1T2. 10. Find P1Z2. 32. (A) What are the odds for rolling a sum of 5 on the single
11. Find P1T ¨ Z2. 12. Find P1R ¨ Z2. roll of two fair dice?
(B) If you bet $1 that a sum of 5 will turn up, what should the
13. Find P1R ∙ Z2. 14. Find P1Z ∙ R2.
house pay (plus return your $1 bet) in order for the game
15. Find P1T ∙ Z2. 16. Are T and Z independent? to be fair?
33. Two coins are flipped 1,000 times with the following fre- In Problems 41–45, urn U1 contains 2 white balls and 3 red balls;
quencies: urn U2 contains 2 white balls and 1 red ball.
41. Two balls are drawn out of urn U1 in succession. What is the
2 heads 210 probability of drawing a white ball followed by a red ball if
1 head 480 the first ball is
0 heads 310 (A) Replaced? (B) Not replaced?
(A) Compute the empirical probability for each outcome. 42. Which of the two parts in Problem 41 involve dependent
events?
(B) Compute the theoretical probability for each outcome.
(C) Using the theoretical probabilities computed in part (B), 43. In Problem 41, what is the expected number of red balls if the
compute the expected frequency of each outcome, as- first ball is
suming fair coins. (A) Replaced? (B) Not replaced?
34. Five dice are rolled all at once. On each of the first four dice, 44. An urn is selected at random by flipping a fair coin; then a
the outcome is a 6. What is the probability that a 6 shows on ball is drawn from the urn. Compute:
the 5th dice? (A) P1R ∙ U1 2 (B) P1R ∙ U2 2
35. An experiment consists of rolling a pair of fair dice. Let X (C) P1U2 ∙ W2 (D) P1U1 ∙ R2
be the random variable associated with the sum of the values 45. In Problem 44, are the events “Selecting urn U1” and “Draw-
that turn up. ing a red ball” independent?
(A) Find the probability distribution for X.
46. From a standard deck of 52 cards, what is the probability of
(B) Find the expected value of X. obtaining a 5-card hand
36. Two dice are rolled. The sample space is chosen as the set of (A) Of all diamonds?
all ordered pairs of integers taken from 51, 2, 3, 4, 5, 66.
(B) Of 3 diamonds and 2 spades?
What is the event A that corresponds to the sum being divis-
ible by 4? What is the event B that corresponds to the sum Write answers in terms of nCr or nPr; do not evaluate.
being divisible by 6? What are P1A2, P1B2, P1A ¨ B2, and 47. In a class of 20 students, two are twins. If 4 students are
P1A ∪ B2? selected at random, what is the probability that the twins are
selected?
37. A person tells you that the following approximate empiri-
cal probabilities apply to the sample space 5e1, e2, e3, e4 6: 48. A 5-card hand is drawn from a standard deck. Discuss how
P1e1 2 ≈ .1, P1e2 2 ≈ - .2, P1e3 2 ≈ .6, P1e4 2 ≈ 2. There you can tell that the following two events are dependent
are three reasons why P cannot be a probability function. without any computation.
Name them.
S = hand consists entirely of spades
38. Use the following information to complete the frequency H = hand consists entirely of hearts
table below:
49. The command in Figure A was used on a graphing calculator
n1A2 = 50, n1B2 = 45,
to simulate 50 repetitions of rolling a pair of dice and record-
n1A ∪ B2 = 80, n1U2 = 100 ing the minimum of the two numbers. A statistical plot of the
A A∙ Totals results is shown in Figure B.
B
B∙
Totals
39. A pointer is spun on a circular spinner. The probabilities of
the pointer landing on the integers from 1 to 5 are given in
the table below.
ei 1 2 3 4 5
(A)
pi .1 .3 .2 .3 .1
(A) What is the probability of the pointer landing on an even 25

number?
(B) What is the probability of the pointer landing on a num-
ber less than 4 given that it landed on an even number?
40. A card is drawn at random from a standard 52-card deck. If E
is the event “The drawn card is red” and F is the event “The
drawn card is an ace,” then 22
0
8
(A) Find P1F 0 E2. (B) Test E and F for independence.

(B)
(A) Use Figure B to find the empirical probability that the 64. Suppose that 3 white balls and 1 black ball are placed in a
minimum is 2. box. Balls are drawn in succession without replacement until
(B) What is the theoretical probability that the minimum is 2? a black ball is drawn, and then the game is over. You win if
the black ball is drawn on the fourth draw.
(C) Using a graphing calculator to simulate 200 rolls of a pair
of dice, determine the empirical probability that the mini- (A) What are the probability and odds for winning?
mum is 4 and compare with the theoretical probability. (B) If you bet $1, what should the house pay you for win-
50. A card is drawn at random from a standard 52-card deck. ning (plus return your $1 bet) if the game is to be fair?
Using a graphing calculator to simulate 800 such draws, de- 65. If each of 4 people is asked to answer a multiple choice ques-
termine the empirical probability that the card is a black jack tion having 7 different options, what is the probability that at
and compare with the theoretical probability. most 3 of them select the same option?
In Problems 51–56, discuss the validity of each statement. 66. Let A and B be events with nonzero probabilities in a sample
If the statement is always true, explain why. If not, give a space S. Under what conditions is P1A ∙ B2 equal to P1B ∙ A2?
counterexample.
51. If P1E2 = 1, then the odds for E are 1 : 1.
52. If E = F′, then P1E ∪ F2 = P1E2 + P1F2.
53. If E and F are complementary events, then E and F are inde-
Applications
pendent. 67. Market research. From a survey of 100 city residents, it
54. If P1E ∪ F2 = 1, then E and F are complementary events. was found that 40 read the daily newspaper, 70 watch the
evening news, and 30 do both. What is the (empirical) prob-
55. If E and F are independent events, then ability that a resident selected at random
P1E2P1F2 = P1E ¨ F2. (A) Reads the daily paper or watches the evening news?
56. If E and F are mutually exclusive events, then (B) Does neither?
P1E2 + P1F2 = P1E ¨ F2.
(C) Does one but not the other?
57. Three fair coins are tossed 1,000 times with the following
frequencies of outcomes: 68. Market research. A market research firm has determined
that 40% of the people in a certain area have seen the ad-
Number of Heads 0 1 2 3 vertising for a new product and that 85% of those who have
Frequency 120 360 350 170 seen the advertising have purchased the product. What is the
probability that a person in this area has seen the advertising
(A) What is the empirical probability of obtaining 2 heads? and purchased the product?
(B) What is the theoretical probability of obtaining 2 heads? 69. Market analysis. A clothing company selected 1,000 per-
(C) What is the expected frequency of obtaining 2 heads? sons at random and surveyed them to determine a relationship
58. You bet a friend $1 that you will get 1 or more double 6’s on 24 between the age of the purchaser and the annual purchases of
rolls of a pair of fair dice. What is your expected value for this jeans. The results are given in the table.
game? What is your friend’s expected value? Is the game fair?
Jeans Purchased Annually
59. If 3 people are selected from a group of 7 men and 3 women, Age 0 1 2 Above 2 Totals
what is the probability that at least 1 woman is selected? Under 12 60 70 30 10 170
Two cards are drawn in succession without replacement from a 12–18 30 100 100 60 290
standard 52-card deck. In Problems 60 and 61, compute the indi- 19–25 70 110 120 30 330
cated probabilities. Over 25 100 50 40 20 210
60. The second card is a heart given that the first card is a heart. Totals 260 330 290 120 1,000
61. The first card is a heart given that the second card is a heart.
62. Two fair (not weighted) dice are each numbered with a 3 on Given the events
one side, a 2 on two sides, and a 1 on three sides. The dice A = person buys 2 pairs of jeans
are rolled, and the numbers on the two up faces are added. B = person is between 12 and 18 years old
If X is the random variable associated with the sample space
S = 52, 3, 4, 5, 66: C = person does not buy more than 2 pairs of jeans
(A) Find the probability distribution of X. D = person buys more than 2 pairs of jeans
(B) Find the expected value of X. (A) Find P1A2, P1B2, P1A ¨ B2, P1A ∙ B2, and P1B ∙ A2.
63. If you pay $3.50 to play the game in Problem 62 (the dice are (B) Are events A and B independent? Explain.
rolled once) and you are returned the dollar amount corre-
sponding to the sum on the faces, what is the expected value (C) Find P1C2, P1D2, P1C ¨ D2, P1C∙D2, and P1D∙C2.
of the game? Is the game fair? If it is not fair, how much (D) Are events C and D mutually exclusive? Independent?
should you pay in order to make the game fair? Explain.
70. Decision analysis. A company sales manager, after careful 74. Medicine: cardiogram test. By testing a large number of in-
analysis, presents two sales plans. It is estimated that plan A dividuals, it has been determined that 82% of the population
will net $10 million if successful (probability .8) and lose $2 have normal hearts, 11% have some minor heart problems,
million if not (probability .2); plan B will net $12 million if and 7% have severe heart problems. Ninety-five percent of
successful (probability .7) and lose $2 million if not (prob- the persons with normal hearts, 30% of those with minor
ability .3). What is the expected return for each plan? Which problems, and 5% of those with severe problems will pass
plan should be chosen based on the expected return? a cardiogram test. What is the probability that a person who
passes the cardiogram test has a normal heart?
71. Insurance. A $2,000 bicycle is insured against theft for an
annual premium of $170. If the probability that the bicycle 75. Genetics. Ten men in 80 and 4 women in 80 are left-handed.
will be stolen during the year is .08 (empirically determined), A person is selected at random and is found to be left-
what is the expected value of the policy? handed. What is the probability that this person is a woman?
(Assume that the ratio of men to women is 5:6 in the total
72. Quality control. Thirteen boxes of drugs, including 3 that
population).
have expired, are sent to a chemist. The chemist will select
three boxes at random and will return the entire shipment if 76. Voter preference. In a straw poll, 30 students in a
1 or more of the sample have expired. What is the probability mathematics class are asked to indicate their preference
that the shipment will be returned? for president of student government. Approximate
empirical probabilities are assigned on the basis of the
73. Quality control. A dozen tablet computers, including 2 that
poll: candidate A should receive 53% of the vote,
are defective, are sent to a computer service center. A random
candidate B should receive 37%, and candidate C
sample of 3 is selected and tested. Let X be the random
should receive 10%. One week later, candidate B wins
variable associated with the number of tablet computers in a
the election. Discuss the factors that may account for
sample that are defective.
the discrepancy between the poll and the election
(A) Find the probability distribution of X. results.
(B) Find the expected number of defective tablet computers
in a sample.
8 Markov Chains
8.1 Properties of Markov

Chains
Introduction
In this chapter, we consider a mathematical model that combines probability
8.2 Regular Markov Chains
and matrices to analyze certain sequences. The model is called a Markov
8.3 Absorbing Markov chain, after the Russian mathematician Andrei Markov (1856–1922). Recent
Chains applications of Markov chains involve a wide variety of topics, including
finance, market research, genetics, medicine, demographics, psychology, and
political science. Problem 92 in Section 8.1, for example, uses a Markov chain
to model a training program for apprentice welders.
In Section 8.1 we introduce the basic properties of Markov chains. In the
remaining sections, we discuss the long-term behavior of two different types of
Markov chains.
436
SECTION 8.1 Properties of Markov Chains 437
8.1 Properties of Markov Chains

■■ Introduction Introduction
■■ Transition and State Matrices In this section, we explore physical systems and their possible states. To understand
■■ Powers of Transition Matrices what this means, consider the following examples:
■■ Application 1. A stock listed on the New York Stock Exchange either increases, decreases, or
does not change in price each day that the exchange is open. The stock can be
thought of as a physical system with three possible states: increase, decrease,
or no change.
2. A commuter, relative to a rapid transit system, can be thought of as a physical
system with two states, a user or a nonuser.
3. During each U.S. congressional election, a voting precinct casts a simple ma-
jority vote for a Republican candidate, a Democratic candidate, or a third-party
candidate. The precinct, relative to all congressional elections past, present,
and future, constitutes a physical system that is in one (and only one) of three
states after each election: Republican, Democratic, or other.
If a system evolves from one state to another in such a way that chance elements are
involved, then the system’s progression through a sequence of states is called a stochastic
process (stochos is the Greek word for “guess”). We will consider a simple example of a
stochastic process, and out of it will arise further definitions and methodology.
A toothpaste company markets a product (brand A) that currently has 10% of the
toothpaste market. The company hires a market research firm to estimate the percent-
age of the market that it might acquire in the future if it launches an aggressive sales
campaign. The research firm uses test marketing and extensive surveys to predict the
effect of the campaign. They find that if a person is using brand A, the probability
is .8 that this person will buy it again when he or she runs out of toothpaste. On the
other hand, a person using another brand will switch to brand A with a probability of
.6 when he or she runs out of toothpaste. So each toothpaste consumer can be consid-
ered to be in one of two possible states:
.2
A = uses brand A or A′ = uses another brand
.8 A A9 .4 The probabilities determined by the market research firm can be represented
.6 graphically in a transition diagram (Fig. 1).
Figure 1 Transition diagram We can also represent this information numerically in a transition probability
matrix:
Next state
A A′
A .8 .2
Current state c d = P
A′ .6 .4

(A) Refer to the transition diagram in Figure 1. What is the probability that a person
using brand A will switch to another brand when he or she runs out of toothpaste?
(B) Refer to transition probability matrix P. What is the probability that a person
who is not using brand A will not switch to brand A when he or she runs out of
toothpaste?
(C) In Figure 1, the sum of the probabilities on the arrows leaving each state is 1.
Will this be true for any transition diagram? Explain your answer.
(D) In transition probability matrix P, the sum of the probabilities in each row is 1.
Will this be true for any transition probability matrix? Explain your answer.
438 CHAPTER 8 Markov Chains
The toothpaste company’s 10% share of the market at the beginning of the sales
campaign can be represented as an initial-state distribution matrix:
A A′
S0 = 3.1 .94
If a person is chosen at random, the probability that this person uses brand A (state A)
is .1, and the probability that this person does not use brand A (state A′) is .9. Thus,
S0 also can be interpreted as an initial-state probability matrix.
What are the probabilities of a person being in state A or A′ on the first purchase
after the start of the sales campaign? Let us look at the probability tree given below.
Note: A0 represents state A at the beginning of the campaign, A′1 represents state A′
on the first purchase after the campaign, and so on.
STATE 0 STATE 1
(INITIAL
STATE)
.8 A1
A0
.1
.2 A91
Start
.6 A1
.9
A90
.4 A91
Proceeding as in Chapter 7, we can read the required probabilities directly from

the tree:
P1A1 2 = P1A0 ¨ A1 2 + P1A0= ¨ A1 2
= 1.121.82 + 1.921.62 = .62
P1A1= 2 = P1A0 ¨ A1= 2 + P1A0= ¨ A1= 2
= 1.121.22 + 1.921.42 = .38
Note: P1A1 2 + P1A′12 = 1, as expected.

The first-state matrix is
A A′
S1 = 3.62 .384
This matrix gives us the probabilities of a randomly chosen person being in state A or
A′ on the first purchase after the start of the campaign. We see that brand A’s market
share has increased from 10% to 62%.
Now, if you were asked to find the probabilities of a person being in state A or
state A′ on the tenth purchase after the start of the campaign, you might start to draw
additional branches on the probability tree. However, you would soon become dis-
couraged because the number of branches doubles for each successive purchase. By
the tenth purchase, there would be 211 = 2,048 branches! Fortunately, we can convert
the summing of branch products to matrix multiplication. In particular, if we multiply
the initial-state matrix S0 by the transition matrix P, we obtain the first-state matrix S1:
A A′ .8 .2 A A′
S0P = 3.1 .94 c d = 31.121.82 + 1.921.62 1.121.22 + 1.921.424 = 3.62 .384 = S1
.6 .4 ¯˚˚˚˘˚˚˚˙ ¯˚˚˚˘˚˚˚˙
Initial First
state Transition state
matrix Compare with the tree
computations above
As you might guess, we can get the second-state matrix S2 (for the second pur-
chase) by multiplying the first-state matrix by the transition matrix:
A A′ A A′
.8 .2
S1P = 3.62 .384 c d = 3.724 .2764 = S2
.6 .4
First Second
state state
The third-state matrix S3 is computed in a similar manner:
A A′ A A′
.8 .2
S2P = 3.724 .2764 c d = 3.7448 .25524 = S3
.6 .4
Second Third
state state
Examining the values in the first three state matrices, we see that brand A’s market
share increases after each toothpaste purchase. Will the market share for brand A con-
tinue to increase until it approaches 100%, or will it level off at some value less than
100%? These questions are answered in Section 8.2 when we develop techniques
for determining the long-run behavior of state matrices.
Transition and State Matrices

The sequence of trials (toothpaste purchases) with the constant transition matrix P is a
special kind of stochastic process called a Markov chain. In general, a Markov chain
is a sequence of experiments, trials, or observations such that the transition probability
matrix from one state to the next is constant. A Markov chain has no memory. The vari-
ous matrices associated with a Markov chain are defined in the next box.
DEFINITION Markov Chains

Given a Markov chain with n states, a kth-state matrix is a matrix of the form
Sk = 3sk1 sk2 g skn 4
such that no entry is negative and the sum of the entries is 1.
Each entry ski is the proportion of the population that is in state i after the kth trial, or,
equivalently, the probability of a randomly selected element of the population being
in state i after the kth trial.
A transition matrix is a constant square matrix P of order n such that the entry in the
ith row and jth column indicates the probability of the system moving from the ith state to
the jth state on the next observation or trial. The sum of the entries in each row must be 1.
CONCEPTUAL INSIGHT
1. Since the entries in a kth-state matrix or transition matrix are probabilities,
they must be real numbers between 0 and 1, inclusive.
2. Rearranging the various states and corresponding transition probabilities in
a transition matrix will produce a different, but equivalent, transition matrix.
For example, both of the following matrices are transition matrices for the
toothpaste company discussed earlier:
A A′ A A′
A .8 .2 A .4 .6
P = c d P′ = c d
A′ .6 .4 A′ .2 .8
Such rearrangements will affect the form of the matrices used in the solution of a
problem but will not affect any of the information obtained from these matrices. In
Section 8.3, we encounter situations where it will be helpful to select a transition
matrix that has a special form. For now, you can choose any order for the states in
a transition matrix.
As we indicated in the preceding discussion, matrix multiplication can be used to

compute the various state matrices of a Markov chain:
If S0 is the initial-state matrix and P is the transition matrix for a
Markov chain, then the subsequent state matrices are given by
S1 = S0P First-state matrix
S2 = S1P Second-state matrix
S3 = S2P Third-state matrix
f
Sk = Sk-1P kth-state matrix
EXAMPLE 1 Insurance An insurance company found that on average, over a period of 10 years,
23% of the drivers in a particular community who were involved in an accident one
year were also involved in an accident the following year. They also found that only
11% of the drivers who were not involved in an accident one year were involved
in an accident the following year. Use these percentages as approximate empirical
probabilities for the following:
(A) Draw a transition diagram.
(B) Find the transition matrix P.
(C) If 5% of the drivers in the community are involved in an accident this year, what
is the probability that a driver chosen at random from the community will be
involved in an accident next year? Year after next?
SOLUTION
(A) .77 A = accident
.23 A A9 .89 A= = no accident
.11
Next year
(B) A A′
This A .23 .77
c d = P Transition matrix
year A′ .11 .89
(C) The initial-state matrix S0 is
A A′
S0 = 3.05 .954 Initial-state matrix
Thus,
A A′ A A′
.23 .77
S0P = 3.05 .954 c d = 3.116 .8844 = S1
.11 .89
This year Next year
(initial state) (first state)
A A′ A A′
.23 .77
S1P = 3.116 .8844 c d = 3.12392 .876084 = S2
.11 .89
Next year Year after next
(first state) (second state)
The probability that a driver chosen at random from the community will have an ac-
cident next year is .116, and the year after next is .12392. That is, it is expected that
11.6% of the drivers in the community will have an accident next year and 12.392%
the year after.
Matched Problem 1 An insurance company classifies drivers as low-risk if

they are accident-free for one year. Past records indicate that 98% of the drivers in
the low-risk category 1L2 one year will remain in that category the next year, and
78% of the drivers who are not in the low-risk category 1L′2 one year will be in
the low-risk category the next year.
(C) If 90% of the drivers in the community are in the low-risk category this year,
what is the probability that a driver chosen at random from the community will
be in the low-risk category next year? Year after next?
Powers of Transition Matrices

Next we investigate the powers of a transition matrix.
The state matrices for a Markov chain are defined recursively; that is, each state
matrix is defined in terms of the preceding state matrix. For example, to find the
fourth-state matrix S4, it is necessary to compute the preceding three state matrices:
S1 = S0P S2 = S1P S3 = S2P S4 = S3P
Is there any way to compute a given state matrix directly without first computing all
the preceding state matrices? If we substitute the equation for S1 into the equation for
S2, substitute this new equation for S2 into the equation for S3, and so on, a definite
pattern emerges:
S1 = S0 P
S2 = S1P = 1S0 P2P = S0 P 2
S3 = S2P = 1S0 P 2 2P = S0 P 3
S4 = S3 P = 1S0 P 3 2P = S0 P 4
f
In general, it can be shown that the kth-state matrix is given by Sk = S0P k. We sum-
marize this important result in Theorem 1.
THEOREM 1 Powers of a Transition Matrix

If P is the transition matrix and S0 is an initial-state matrix for a Markov chain,
then the kth-state matrix is given by
Sk = S0 P k
The entry in the ith row and jth column of P k indicates the probability of the system
moving from the ith state to the jth state in k observations or trials. The sum of the
entries in each row of P k is 1.
EXAMPLE 2 Using P k to Compute Sk Find P 4 and use it to find S4 for

A A′
A .1 .9 A A′
P = c d and S0 = 3.2 .84
A′ .6 .4
.1 .9 .1 .9 .55 .45
SOLUTION P 2 = PP = c dc d = c d
.6 .4 .6 .4 .3 .7
.55 .45 .55 .45 .4375 .5625
P 4 = P 2P 2 = c dc d = c d
.3 .7 .3 .7 .375 .625
.4375 .5625
S4 = S0P 4 = 3.2 .84 c d = 3.3875 .61254
.375 .625
Matched Problem 2 Find P 4 and use it to find S4 for

A A′
A .8 .2 A A′
P = c d and S0 = 3.8 .24
A′ .3 .7
If a graphing calculator or a computer is available for computing matrix
products and powers of a matrix, finding state matrices for any number of trials
becomes a routine calculation.
EXAMPLE 3 Using a Graphing Calculator and P k to Compute S k Use P 8 and a graphing

calculator to find S8 for P and S0 as given in Example 2. Round values in S8 to six
decimal places.
SOLUTION After storing the matrices P and S0 in the graphing calculator’s memory,
we use the equation
S8 = S0 P 8
to compute S8. Figure 2 shows the result on a typical graphing calculator, where the matrix
names [A] and [B] denote P and S0, respectively. We see that (to six decimal places)
S8 = 3.399219 .6007814
Figure 2
Matched Problem 3 Use P 8 and a graphing calculator to find S8 for P and S0
as given in Matched Problem 2. Round values in S8 to six decimal places.
Application
The next example illustrates the use of Theorem 1 in an application.
EXAMPLE 4 Student Retention Part-time students in a university MBA program are consid-
ered to be entry-level students until they complete 15 credits successfully. Then they
are classified as advanced-level students and can take more advanced courses and
work on the thesis required for graduation. Past records indicate that at the end of
each year, 10% of the entry-level students (E) drop out of the program (D) and 30%
become advanced-level students (A). Also, 10% of the advanced-level students drop
out of the program and 40% graduate (G) each year. Students that graduate or drop
out never return to the program.
(A) Draw a transition diagram. (B) Find the transition matrix P.
(C) What is the probability that an entry-level student graduates within 4 years?
Drops out within 4 years?
SOLUTION
(A) If 10% of entry-level students drop out and 30% become advanced-level
students, then the remaining 60% must continue as entry-level students for
another year (see the diagram). Similarly, 50% of advanced-level students
must continue as advanced-level students for another year. Since students
who drop out never return, all students in state D in one year will continue
in that state the next year. We indicate this by placing a 1 on the arrow from
D back to D. State G is labeled in the same manner.
.1
.6 E D 1
.3 Transition diagram
.1
.5 A G 1
.4
E D A G
(B) E .6 .1 .3 0
D 0 1 0 0
P = D T Transition matrix
A 0 .1 .5 .4
G 0 0 0 1
(C) The probability that an entry-level student moves from state E to state G within
4 years is the entry in row 1 and column 4 of P 4 (Theorem 1). Hand computation
of P 4 requires two multiplications:
.6 .1 .3 0 .6 .1 .3 0 .36 .19 .33 .12

0 1 0 0 0 1 0 0 0 1 0 0
P2 = D TD T = D T
0 .1 .5 .4 0 .1 .5 .4 0 .15 .25 .6
0 0 0 1 0 0 0 1 0 0 0 1
.36 .19 .33 .12 .36 .19 .33 .12

0 1 0 0 0 1 0 0
P 4 = P 2P 2 = D TD T
0 .15 .25 .6 0 .15 .25 .6
0 0 0 1 0 0 0 1
.1296 .3079 .2013 .3612

0 1 0 0
= D T
0 .1875 .0625 .75
0 0 0 1
The probability that an entry-level student has graduated within 4 years is .3612.
Similarly, the probability that an entry-level student has dropped out within 4 years
is .3079 (the entry in row 1 and column 2 of P 4).
Matched Problem 4 Refer to Example 4. At the end of each year the faculty
examines the progress that each advanced-level student has made on the required
thesis. Past records indicate that 30% of advanced-level students (A) complete the
thesis requirement (C) and 10% are dropped from the program for insufficient pro-
gress (D), never to return. The remaining students continue to work on their theses.
(C) What is the probability that an advanced-level student completes the thesis
requirement within 4 years? Is dropped from the program for insufficient prog-
ress within 4 years?

Refer to Example 4. States D and G are referred to as absorbing states because a
student who enters either one of these states never leaves it. Absorbing states are dis-
cussed in detail in Section 8.3.
(A) How can absorbing states be recognized from a transition diagram? Draw a
transition diagram with two states, one that is absorbing and one that is not, to
illustrate.
(B) How can absorbing states be recognized from a transition matrix? Write the
transition matrix for the diagram you drew in part (A) to illustrate.
Exercises 8.1
Skills Warm-up Exercises In Problems 27–32, use the transition diagram
W In Problems 1–8, find the matrix product, if it is defined. (If neces-
.3
sary, review Section 3.4.) B .8
2 5 3 6 9 .1
1. c dc d 2. 34 54 c d .2 A .1 .2
4 1 2 3 7
.4
3 2 5 6 9
3. c d c d 4. c d 34 54 .5
C .4
2 4 1 3 7
2 5 4 6 9 to find S1 and S2 for the indicated initial state matrix S0.

5. 33 24 c d 6. c d c d
4 1 5 3 7 27. S0 = 31 0 04 28. S0 = 30 1 04
2 5 6 9 4 29. S0 = 30 .4 .64 30. S0 = 3.8 0 .24
7. c d 33 24 8. c dc d
4 1 3 7 5 31. S0 = 3.5 .2 .34 32. S0 = [.2 .7 .1]
33. Draw the transition diagram that corresponds to the transition
A In Problems 9–14, use the transition matrix
matrix of Problem 9.
A B
A .7 .3 34. Find the transition matrix that corresponds to the transition
P = c d diagram of Problem 15.
B .1 .9
35. Draw the transition matrix that corresponds to the transition
to find S1 and S2 for the indicated initial state matrix S0. diagram of Problem 27.
9. S0 = 30 14 10. S0 = 31 04 36. Find the transition diagram that corresponds to the transition
matrix of Problem 21.
11. S0 = 3.6 .44 12. S0 = 3.2 .84
13. S0 = 3.25 .754 14. S0 = 3.75 .254 In Problems 37–44, could the given matrix be the transition
matrix of a Markov chain?
In Problems 15–20, use the transition diagram
.5 .3 .7 .9 .1
37. c d 38. c d
.5 A B .2 1 0 .4 .8
.8 .5 .5 0 1
39. c d 40. c d
.7 - .3 1 0
to find S1 and S2 for the indicated initial state matrix S0.
.2 .8
15. S0 = 31 04 16. S0 = 30 14 .1 .3 .6
41. c d 42. C .5 .5 S
.2 .4 .4
17. S0 = 3.3 .74 18. S0 = 3.9 .14 .9 .1
19. S0 = 3.5 .54 20. S0 = 3.2 .84 .5 .1 .4 .3 .3 .4
43. C 0 .5 .5 S 44. C .7 .2 .2 S
In Problems 21–26, use the transition matrix
.2 .1 .7 .1 .8 .1
A B C
A .2 .4 .4
In Problems 45–50, is there a unique way of filling in the missing
P = B C .7 .2 .1 S
probabilities in the transition diagram? If so, complete the transi-
C .5 .3 .2 tion diagram and write the corresponding transition matrix. If not,
explain why.
to find S1 and S2 for the indicated initial state matrix S0.
21. S0 = 30 1 04 22. S0 = 30 0 14 45. ?
.4 A B ?
23. S0 = 3.5 0 .54 24. S0 = 3.5 .5 04
.7
25. S0 = 3.1 .3 .64 26. S0 = 3.4 .3 .34
46. .9
? A B .2
?
47. .6 is .2, and the probability of going from state B to state C in

.4 A B ? one trial is .5. The probability of going from state C to state C
in one trial is 1.
?
60. A Markov chain has three states, A, B, and C. The probability
48. ?
of going from state A to state B in one trial is 1. The probability
? A B .7 of going from state B to state A in one trial is .5, and the prob-
.3 ability of going from state B to state C in one trial is .5. The
probability of going from state C to state A in one trial is 1.
49.
.4
B .2 Problems 61–70 refer to the following transition matrix P and its
.5
powers:
? A ? ? A B C A B C
.7 A
.6 .3 .1 A .43 .35 .22
C .1 P = B C .2 .5 .3 S P 2 = B C .25 .37 .38 S
.5
C .1 .2 .7 C .17 .27 .56
50. A B C
?
B .2 A .35 .348 .302
?
P 3 = B C .262 .336 .402 S
.3 A .2 .8 C .212 .298 .49
.1
61. Find the probability of going from state A to state B in two trials.
C ?
.4 62. Find the probability of going from state B to state C in two trials.
63. Find the probability of going from state C to state A in three trials.
In Problems 51–56, are there unique values of a, b, and c that make
P a transition matrix? If so, complete the transition matrix and 64. Find the probability of going from state B to state B in three trials.
draw the corresponding transition diagram. If not, explain why. 65. Find S2 for S0 = [1 0 0] and explain what it represents.
A B C A B C 66. Find S2 for S0 = [0 1 0] and explain what it represents.
A 0 .5 a A a 0 .9
51. P = B C b 0 .4 S 52. P = B C .2 .3 b S 67. Find S3 for S0 = [0 0 1] and explain what it represents.
C .2 c .1 C .6 c 0 68. Find S3 for S0 = [1 0 0] and explain what it represents.
A B C A B C 69. Using a graphing calculator to compute powers of P, find the

A 0 a .3 A 0 1 a smallest positive integer n such that the corresponding entries
53. P = B C 0 b 0S 54. P = B C 0 0 bS in P n and P n + 1 are equal when rounded to two decimal places.
C c .8 0 C c .5 0 70. Using a graphing calculator to compute powers of P, find the
smallest positive integer n such that the corresponding entries in
A B C A B C P n and P n + 1 are equal when rounded to three decimal places.
A
.2 .1 .7 A a .8 .1
In Problems 71–74, given the transition matrix P and initial-state
55. P = B C a .4 cS 56. P = B C .3 b .4 S
matrix S0, find P 4 and use P 4 to find S4.
C .5 b .4 C .6 .5 c
A B
B In Problems 57–60, use the given information to draw the transi- 71. P = A c .1 .9
d ; S0 = 3.8 .24
tion diagram and find the transition matrix. B .6 .4
A B
57. A Markov chain has two states, A and B. The probability of A .8 .2
going from state A to state B in one trial is .7, and the prob- 72. P = c d ; S0 = 3.4 .64
B .3 .7
ability of going from state B to state A in one trial is .9.
A B C
58. A Markov chain has two states, A and B. The probability of A 0 .4 .6
going from state A to state A in one trial is .6, and the prob- 73. P = B C 0 0 1 S ; S0 = 3.2 .3 .54
ability of going from state B to state B in one trial is .2.
C 1 0 0
59. A Markov chain has three states, A, B, and C. The probabil- A B C
ity of going from state A to state B in one trial is .1, and the A 0 1 0
probability of going from state A to state C in one trial is .3.
74. P = B C .8 0 .2 S ; S0 = 3.4 .2 .44
The probability of going from state B to state A in one trial
C 1 0 0
75. A Markov chain with two states has transition matrix P. If the .9 .1
initial-state matrix is S0 = [1 0], discuss the relationship 82. Repeat Problem 81 for P = c d.
.4 .6
between the entries in the kth-state matrix and the entries in
the kth power of P. 83. Refer to Problem 81. Find P k for k = 2, 4, 8, c. Can you
identify a matrix Q that the matrices P k are approaching? If so,
76. Repeat Problem 75 if the initial-state matrix is S0 = [0 1].
how is Q related to the results you discovered in Problem 81?
C 77. Given the transition matrix
84. Refer to Problem 82. Find P k for k = 2, 4, 8, c . Can you
A B C D identify a matrix Q that the matrices P k are approaching? If so,
A .2 .2 .3 .3 how is Q related to the results you discovered in Problem 82?
B 0 1 0 0
P = D T
C .2 .2 .1 .5
D 0 0 0 1
(A) Find P 4. Applications
(B) Find the probability of going from state A to state D in 85. Scheduling. An outdoor restaurant in a summer resort closes
four trials. only on rainy days. From past records, it is found that from
(C) Find the probability of going from state C to state B in May through September, when it rains one day, the prob-
four trials. ability of rain for the next day is .4; when it does not rain one
day, the probability of rain for the next day is .06.
(D) Find the probability of going from state B to state A in
four trials. (A) Draw a transition diagram.
78. Repeat Problem 77 for the transition matrix (B) Write the transition matrix.
A B C D
(C) If it rains on Thursday, what is the probability that the
A .5 .3 .1 .1
restaurant will be closed on Saturday? On Sunday?
B 0 1 0 0
P = D T 86. Scheduling. Repeat Problem 85 if the probability of rain fol-
C 0 0 1 0
lowing a rainy day is .6 and the probability of rain following
D .1 .2 .3 .4
a nonrainy day is .1.
A matrix is called a probability matrix if all its entries are real
87. Advertising. A television advertising campaign is conducted
numbers between 0 and 1, inclusive, and the sum of the entries in
during the football season to promote a well-known brand X
each row is 1. So transition matrices are square probability matri-
shaving cream. For each of several weeks, a survey is made,
ces and state matrices are probability matrices with one row.
and it is found that each week, 80% of those using brand X
79. Show that if continue to use it and 20% switch to another brand. It is also
found that of those not using brand X, 20% switch to brand X
a 1 - a while the other 80% continue using another brand.
P = c d
1 - b b
is a probability matrix, then P 2 is a probability matrix.
(B) Write the transition matrix.
80. Show that if
(C) If 20% of the people are using brand X at the start of the
a 1 - a advertising campaign, what percentage will be using it
P = c d and S = 3c 1 - c4
1 - b b 1 week later? 2 weeks later?
are probability matrices, then SP is a probability matrix. 88. Car rental. A car rental agency has facilities at both JFK
k
Use a graphing calculator and the formula Sk = S0 P (Theorem 1) and LaGuardia airports. Assume that a car rented at either
to compute the required state matrices in Problems 81–84. airport must be returned to one or the other airport. If a car is
rented at LaGuardia, the probability that it will be returned
81. The transition matrix for a Markov chain is there is .8; if a car is rented at JFK, the probability that it will
.4 .6 be returned there is .7. Assume that the company rents all its
P = c d 100 cars each day and that each car is rented (and returned)
.2 .8
only once a day. If we start with 50 cars at each airport, then
(A) If S0 = 30 14, find S2, S4, S8, c. Can you identify a
state matrix S that the matrices Sk seem to be approaching? (A) What is the expected distribution on the next day?
(B) Repeat part (A) for S0 = 31 04. (B) What is the expected distribution 2 days later?
(C) Repeat part (A) for S0 = 3.5 .54. 89. Homeowner’s insurance. In a given city, the market for
homeowner’s insurance is dominated by two companies:
(D) Find SP for any matrix S you identified in parts (A)–(C). National Property and United Family. Currently, National
(E) Write a brief verbal description of the long-term behav- Property insures 50% of homes in the city, United Family
ior of the state matrices of this Markov chain based on insures 30%, and the remainder are insured by a collection
your observations in parts (A)–(D). of smaller companies. United Family decides to offer rebates
to increase its market share. This has the following effects on terminated for unsatisfactory performance, and the remainder
insurance purchases for the next several years: each year 25% continue as apprentices.
of National Property’s customers switch to United Family
and 10% switch to other companies; 10% of United Family’s
customers switch to National Property and 5% switch to other (B) Write the transition matrix.
companies; 15% of the customers of other companies switch
(C) What is the probability that an apprentice is promoted to
to National Property and 35% switch to United Family.
professional welder within 2 years? Within 4 years?
93. Health plans. A midwestern university offers its employees
(B) Write the transition matrix. three choices for health care: a clinic-based health maintenance
organization (HMO), a preferred provider organization (PPO),
(C) What percentage of homes will be insured by National
and a traditional fee-for-service program (FFS). Each year,
Property next year? The year after next?
the university designates an open enrollment period during
(D) What percentage of homes will be insured by United which employees may change from one health plan to another.
Family next year? The year after next? Prior to the last open enrollment period, 20% of employees
were enrolled in the HMO, 25% in the PPO, and the remain-
90. Service contracts. A small community has two heating
der in the FFS. During the open enrollment period, 15% of
services that offer annual service contracts for home heating:
employees in the HMO switched to the PPO and 5% switched
Alpine Heating and Badger Furnaces. Currently, 25% of home-
to the FFS, 20% of the employees in the PPO switched to the
owners have service contracts with Alpine, 30% have service
HMO and 10% to the FFS, and 25% of the employees in the
contracts with Badger, and the remainder do not have service
FFS switched to the HMO and 30% switched to the PPO.
contracts. Both companies launch aggressive advertising cam-
paigns to attract new customers, with the following effects on (A) Write the transition matrix.
service contract purchases for the next several years: each year
(B) What percentage of employees were enrolled in each
35% of homeowners with no current service contract decide to
health plan after the last open enrollment period?
purchase a contract from Alpine and 40% decide to purchase
one from Badger. In addition, 10% of the previous customers (C) If this trend continues, what percentage of employees
at each company decide to switch to the other company, and will be enrolled in each plan after the next open
5% decide they do not want a service contract. enrollment period?
(A) Draw a transition diagram. 94. Dental insurance. Refer to Problem 93. During the open
enrollment period, university employees can switch between
two available dental care programs: the low-option plan
(C) What percentage of homes will have service contracts (LOP) and the high-option plan (HOP). Prior to the last
with Alpine next year? The year after next? open enrollment period, 40% of employees were enrolled in
the LOP and 60% in the HOP. During the open enrollment
(D) What percentage of homes will have service contracts
program, 30% of employees in the LOP switched to the HOP
with Badger next year? The year after next?
and 10% of employees in the HOP switched to the LOP.
91. Travel agent training. A chain of travel agencies main-
(A) Write the transition matrix.
tains a training program for new travel agents. Initially, all
new employees are classified as beginning agents requiring (B) What percentage of employees were enrolled in each
extensive supervision. Every 6 months, the performance of dental plan after the last open enrollment period?
each agent is reviewed. Past records indicate that after each
(C) If this trend continues, what percentage of employees
semiannual review, 40% of the beginning agents are promot-
will be enrolled in each dental plan after the next open
ed to intermediate agents requiring only minimal supervision,
enrollment period?
10% are terminated for unsatisfactory performance, and the
remainder continue as beginning agents. Furthermore, 30% 95. Housing trends. The 2000 census reported that 41.9% of
of the intermediate agents are promoted to qualified travel the households in the District of Columbia were homeown-
agents requiring no supervision, 10% are terminated for ers and the remainder were renters. During the next decade,
unsatisfactory performance, and the remainder continue as 15.3% of homeowners became renters, and the rest continued
intermediate agents. to be homeowners. Similarly, 17.4% of renters became
homeowners, and the rest continued to rent.
(A) Write the appropriate transition matrix.
(B) According to this transition matrix, what percentage of
(C) What is the probability that a beginning agent is pro-
households were homeowners in 2010?
moted to qualified agent within 1 year? Within 2 years?
(C) If the transition matrix remains the same, what percent-
92. Welder training. All welders in a factory begin as ap-
age of households will be homeowners in 2030?
prentices. Every year the performance of each apprentice is
reviewed. Past records indicate that after each review, 10% of 96. Housing trends. The 2000 census reported that 66.4% of the
the apprentices are promoted to professional welder, 20% are households in Alaska were homeowners, and the remainder
were renters. During the next decade, 37.2% of the home- .625 .375
owners became renters, and the rest continued to be home- 2. P 4 = c d ; S4 = 3.6125 .38754
.5625 .4375
owners. Similarly, 71.5% of the renters became homeowners,
and the rest continued to rent. 3. S8 = 3.600781 .3992194
(A) Write the appropriate transition matrix. 4. (A)

.1 D 1
(B) According to this transition matrix, what percentage of
households were homeowners in 2010?
.6 A
(C) If the transition matrix remains the same, what percent-
age of households will be homeowners in 2030?
.3 C 1
1. (A) .02 L = Low@risk A C D
L= = Not low@risk A .6 .3 .1
.98 L L9 .22
(B) P = C C 0 1 0S
.78
D 0 0 1
(B) Next year (C) Next year: .96;
L L′ year after next: .972 (C) .6528; .2176
This L .98 .02
c d = P
year L′ .78 .22

■■ Stationary Matrices Given a Markov chain with transition matrix P and initial-state matrix S0, the entries
■■ Regular Markov Chains
in the state matrix Sk are the probabilities of being in the corresponding states after
k trials. What happens to these probabilities as the number of trials k increases? In
■■ Applications this section, we establish conditions on the transition matrix P that enable us to
■■ Graphing Calculator determine the long-run behavior of both the state matrices Sk and the powers of the
Approximations transition matrix P k.
Stationary Matrices
We begin by considering a concrete example—the toothpaste company discussed
earlier. Recall that the transition matrix was given by
A A′
A .8 .2 A = uses brand A toothpaste
P = c d
A′ .6 .4 A′ = uses another brand
Initially, this company had a 10% share of the toothpaste market. If the probabilities
in the transition matrix P remain valid over a long period of time, what will happen
to the company’s market share? Examining the first several state matrices will give us
some insight into this situation (matrix multiplication details are omitted):
S0 = 3.1 .94
S1 = S0 P = 3.62 .384
S2 = S1P = 3.724 .2764
S3 = S2P = 3.7448 .25524
S4 = S3P = 3.74896 .251044
S5 = S4P = 3.749792 .2502084
S6 = S5P = 3.7499584 .25004164
SECTION 8.2 Regular Markov Chains 449
It appears that the state matrices are getting closer and closer to S = [.75 .25] as
we proceed to higher states. Let us multiply the matrix S (the matrix that the other
state matrices appear to be approaching) by the transition matrix:
.8 .2
SP = 3.75 .254 c d = 3.75 .254 = S
.6 .4
No change occurs! The matrix 3.75 .254 is called a stationary matrix. If we reach
this state or are very close to it, the system is said to be at a steady state; that is, later
states either will not change or will not change very much. In terms of this example,
this means that in the long run a person will purchase brand A with a probability of
.75. In other words, the company can expect to capture 75% of the market, assuming
that the transition matrix does not change.
DEFINITION Stationary Matrix for a Markov Chain

The state matrix S = [s1 s2 c sn] is a stationary matrix for a Markov chain
with transition matrix P if
SP = S
where si Ú 0, i = 1, c , n, and s1 + s2 + g + sn = 1.

(A) Suppose that the toothpaste company started with only 5% of the market instead
of 10%. Write the initial-state matrix and find the next six state matrices. Dis-
cuss the behavior of these state matrices as you proceed to higher states.
(B) Repeat part (A) if the company started with 90% of the toothpaste market.
Regular Markov Chains

Does every Markov chain have a unique stationary matrix? And if a Markov chain
has a unique stationary matrix, will the successive state matrices always approach
this stationary matrix? Unfortunately, the answer to both these questions is no (see
Problems 47–50, Exercises 8.2). However, there is one important type of Markov
chain for which both questions always can be answered in the affirmative. These are
called regular Markov chains.
DEFINITION Regular Markov Chains

A transition matrix P is regular if some power of P has only positive entries.
A Markov chain is a regular Markov chain if its transition matrix is regular.
EXAMPLE 1 Recognizing Regular Matrices Which of the following matrices are regular?
.5 .5 0
.8 .2 0 1
(A) P = c d (B) P = c d (C) P = C 0 .5 .5 S
.6 .4 1 0
1 0 0
SOLUTION
(A) This is the transition matrix for the toothpaste company. Since all the entries in
P are positive, we can immediately conclude that P is regular.
(B) P has two 0 entries, so we must examine higher powers of P:

1 0 0 1 1 0 0 1
P2 = c d P3 = c d P4 = c d P5 = c d
0 1 1 0 0 1 1 0
Since the powers of P oscillate between P and I, the 2 * 2 identity, all powers
of P will contain 0 entries. Hence, P is not regular.
(C) Again, we examine higher powers of P:
.25 .5 .25 .375 .375 .25
2 3
P = C .5 .25 .25 S P = C .5 .375 .125 S
.5 .5 0 .25 .5 .25
Since all the entries in P 3 are positive, P is regular.
Matched Problem 1 Which of the following matrices are regular?

0 1 0
.3 .7 1 0
(A) P = c d (B) P = c d (C) P = C .5 0 .5 S
1 0 1 0
.5 0 .5
The relationships among successive state matrices, powers of the transition ma-
trix, and the stationary matrix for a regular Markov chain are given in Theorem 1.
The proof of this theorem is left to more advanced courses.
THEOREM 1 Properties of Regular Markov Chains

Let P be the transition matrix for a regular Markov chain.
(A) There is a unique stationary matrix S that can be found by solving the equation
SP = S
(B) Given any initial-state matrix S0, the state matrices Sk approach the stationary
matrix S.
(C) The matrices P k approach a limiting matrix P, where each row of P is equal
to the stationary matrix S.
EXAMPLE 2 Finding the Stationary Matrix The transition matrix for a Markov chain is
.7 .3
P = c d
.2 .8
(A) Find the stationary matrix S.
(B) Discuss the long-run behavior of Sk and P k.
SOLUTION
(A) Since P is regular, the stationary matrix S must exist. To find it, we must solve
the equation SP = S. Let
S = 3s1 s2 4
and write
.7 .3
3s1 s2 4 c d = 3s1 s2 4
.2 .8
After multiplying the left side, we obtain

31.7s1 + .2s2 2 1.3s1 + .8s2 24 = 3s1 s2 4
which is equivalent to the system
.7s1 + .2s2 = s1 or - .3s1 + .2s2 = 0
(1)
.3s1 + .8s2 = s2 or .3s1 - .2s2 = 0
System (1) is dependent and has an infinite number of solutions. However, we
are looking for a solution that is also a state matrix. This gives us another equa-
tion that we can add to system (1) to obtain a system with a unique solution.
- .3s1 + .2s2 = 0
.3s1 - .2s2 = 0 (2)
s1 + s2 = 1
System (2) can be solved using matrix methods or elimination to obtain
s1 = .4 and s2 = .6
Therefore,
S = 3.4 .64
is the stationary matrix.
CHECK
.7 .3
SP = 3.4 .64 c d = 3.4 .64 = S
.2 .8
(B) Given any initial-state matrix S0, Theorem 1 guarantees that the state matrices
Sk will approach the stationary matrix S. Furthermore,
.7 .3 k .4 .6
Pk = c d approaches the limiting matrix P = c d
.2 .8 .4 .6
Matched Problem 2 The transition matrix for a Markov chain is

.6 .4
P = c d
.1 .9
Find the stationary matrix S and the limiting matrix P.
Applications
EXAMPLE 3 Insurance Refer to Example 1 in Section 8.1, where we found the following
transition matrix for an insurance company:
A A′
A .23 .77 A = accident
P = c d
A′ .11 .89 A′ = no accident
If these probabilities remain valid over a long period of time, what percentage of
drivers are expected to have an accident during any given year?
SOLUTION To determine what happens in the long run, we find the stationary
matrix by solving the following system:
.23 .77
3s1 s2 4 c d = 3s1 s2 4 and s1 + s2 = 1
.11 .89
which is equivalent to
.23s1 + .11s2 = s1 or - .77s1 + .11s2 = 0
.77s1 + .89s2 = s2 .77s1 - .11s2 = 0
s1 + s2 = 1 s1 + s2 = 1
Solving this system, we obtain
s1 = .125 and s2 = .875
The stationary matrix is 3.125 .8754, which means that in the long run, assuming
that the transition matrix does not change, about 12.5% of drivers in the community
will have an accident during any given year.
Matched Problem 3 Refer to Matched Problem 1 in Section 8.1, where we

found the following transition matrix for an insurance company:
L L′
L .98 .02 L = low-risk
P = c d
L′ .78 .22 L′ = not low-risk
If these probabilities remain valid for a long period of time, what percentage of
drivers are expected to be in the low-risk category during any given year?
EXAMPLE 4 Employee Evaluation A company rates every employee as below average, av-
erage, or above average. Past performance indicates that each year, 10% of the
below-average employees will raise their rating to average, and 25% of the average
employees will raise their rating to above average. On the other hand, 15% of the
average employees will lower their rating to below average, and 15% of the above-
average employees will lower their rating to average. Company policy prohibits
rating changes from below average to above average, or conversely, in a single year.
Over the long run, what percentage of employees will receive below-average rat-
ings? Average ratings? Above-average ratings?
SOLUTION First we find the transition matrix:
Next year
-
A A A+
A- .9 .1 0 A- = below average
This A C .15 .6 .25 S A = average
year
A+ 0 .15 .85 A+ = above average
To determine what happens over the long run, we find the stationary matrix by solv-
ing the following system:
.9 .1 0
3s1 s2 s3 4 C .15 .6 .25 S = 3s1 s2 s3 4 and s1 + s2 + s3 = 1
0 .15 .85
which is equivalent to
.9s1 + .15s2 = s1 or - .1s1 + .15s2 = 0
.1s1 + .6s2 + .15s3 = s2 .1s1 - .4s2 + .15s3 = 0
.25s2 + .85s3 = s3 .25s2 - .15s3 = 0
s1 + s2 + s3 = 1 s1 + s2 + s3 = 1
Using Gauss–Jordan elimination to solve this system of four equations with three
variables, we obtain
s1 = .36 s2 = .24 s3 = .4
In the long run, 36% of employees will be rated as below average, 24% as average,
and 40% as above average.
Matched Problem 4 A mail-order company classifies its customers as pre-

ferred, standard, or infrequent depending on the number of orders placed in a year.
Past records indicate that each year, 5% of preferred customers are reclassified
as standard and 12% as infrequent, 5% of standard customers are reclassified as
preferred and 5% as infrequent, and 9% of infrequent customers are reclassified
as preferred and 10% as standard. Assuming that these percentages remain valid,
what percentage of customers are expected to be in each category in the long run?
Graphing Calculator Approximations

If P is the transition matrix for a regular Markov chain, then the powers of P approach the
limiting matrix P, where each row of P is equal to the stationary matrix S (Theorem 1C).
We can use this result to approximate S by computing P k for sufficiently large values of k.
The next example illustrates this approach on a graphing calculator.
EXAMPLE 5 Approximating the Stationary Matrix Compute powers of the transition matrix
P to approximate P and S to four decimal places. Check the approximation in the
equation SP = S.
.5 .2 .3
P = C .7 .1 .2 S
.4 .1 .5
SOLUTION To approximate P to four decimal places, we store P in a graphing cal-

culator using the matrix name [A] (Fig. 1A), set the decimal display to four places,
and compute powers of P until all three rows of P k are identical. Examining the
output in Figure 1B, we conclude that
.4943 .1494 .3563

P = C .4943 .1494 .3563 S and S = 3.4943 .1494 .35634
.4943 .1494 .3563
Entering S in the graphing calculator using the matrix name [B], and computing SP
shows that these matrices are correct to four decimal places (Fig. 1C).
(A) P (B) P 9 (C) Check: SP 5 S

Figure 1
.3 .6 .1
P = C .2 .3 .5 S
.1 .2 .7
CONCEPTUAL INSIGHT
1. We used a relatively small value of k to approximate P in Example 5. Many
graphing calculators will compute P k for large values of k almost as rapidly as
for small values. However, round-off errors can occur in these calculations. A
safe procedure is to start with a relatively small value of k, such as k = 8, and
then keep doubling k until the rows of P k are identical to the specified number
of decimal places.
2. If any of the entries of P k are approaching 0, then the graphing calculator
may use scientific notation to display these entries as very small numbers.
Figure 2 shows the 100th power of a transition matrix P using the matrix name
[A]. The entry in row 2 and column 2 of P 100 is approaching 0, but the graph-
ing calculator displays it as 5.1538 * 10 -53. If this occurs, simply change this
Figure 2 value to 0 in the corresponding entry in P. Thus, from the output in Figure 2 we
conclude that
1 0 k 1 0
Pk = c d approaches P = c d
.7 .3 1 0
Exercises 8.2
Skills Warm-up Exercises .1 .9 .5 .5
11. J R 12. J R
In Problems 1–8, without using a calculator, find P . (If100 .5 .4 .8 .2
necessary,review Section 3.4.) [Hint: First find P 2.]
.4 .6 .4 .6
13. J R 14. J R
1 0 0 0 0 1 1 0
1. c d 2. c d
0 0 1 0
0 1 .3 .7
1 0 0 1 15. J R 16. J R
3. c d 4. c d .8 .2 .2 .6
0 1 1 0
.6 .4
1 0 0 1 0 0 .2 .5 .3
17. C .1 .9 S 18. J R
5. C 0 0 0S 6. C 0 1 0S .6 .3 .1
.3 .7
0 0 1 0 0 1
0 1 0 .2 0 .8
0 0 1 0 1 1 19. C 0 0 1S 20. C 0 0 1S
7. C 0 0 0S 8. C 0 0 1S .5 .5 0 .7 0 .3
0 0 0 0 0 0 .1 .3 .6 0 0 1
21. C .8 .1 .1 S 22. C .9 0 .1 S
A In Problems 9–22, could the given matrix be the transition matrix 0 0 1 0 1 0
of a regular Markov chain?
.6 .4 .3 .7
9. J R 10. J R
.4 .6 .2 .6
B For each transition matrix P in Problems 23–30, solve the equation .5 .5 0 .2 .2 .6

SP = S to find the stationary matrix S and the limiting matrix P. 43. P = C 0 .5 .5 S 44. P = C .5 0 .5 S
.1 .9 .8 .2 .8 .1 .1 .5 0 .5
23. P = c d 24. P = c d
.6 .4 .3 .7
C 45. A red urn contains 2 red marbles and 3 blue marbles, and a
blue urn contains 1 red marble and 4 blue marbles. A marble
.5 .5 .9 .1
25. P = c d 26. P = c d is selected from an urn, the color is noted, and the marble is
.3 .7 .7 .3 returned to the urn from which it was drawn. The next marble
.5 .1 .4 .4 .1 .5 is drawn from the urn whose color is the same as the marble
just drawn. This is a Markov process with two states: draw
27. P = C .3 .7 0S 28. P = C .2 .8 0S
from the red urn or draw from the blue urn.
0 .6 .4 0 .5 .5
(A) Draw a transition diagram for this process.
.8 .2 0 .2 .8 0 (B) Write the transition matrix.
29. P = C .5 .1 .4 S 30. P = C .6 .1 .3 S (C) Find the stationary matrix and describe the long-run
0 .6 .4 0 .9 .1 behavior of this process.
46. Repeat Problem 45 if the red urn contains 5 red and 3 blue
Problems 31–34 refer to the regular Markov chain with transition marbles, and the blue urn contains 1 red and 3 blue marbles.
matrix
47. Given the transition matrix
.5 .5
P = c d 0 1
.2 .8 P = c d
1 0
31. For S = 3.2 .54, calculate SP. Is S a stationary matrix? (A) Discuss the behavior of the state matrices S1, S2, S3, c
Explain. for the initial-state matrix S0 = 3.2 .84.
32. For S = 3.6 1.54, calculate SP. Is S a stationary matrix? (B) Repeat part (A) for S0 = 3.5 .54.
Explain.
(C) Discuss the behavior of P , k = 2, 3, 4, c.
k
33. For S = 30 04, calculate SP. Is S a stationary matrix?
Explain. (D) Which of the conclusions of Theorem 1 are not valid for
2 5 this matrix? Why is this not a contradiction?
34. For S = c d , calculate SP. Is S a stationary matrix? 48. Given the transition matrix
7 7
Explain.
0 1 0
In Problems 35–40, discuss the validity of each statement. If the P = C0 0 1S
statement is always true, explain why. If not, give a counterexample. 1 0 0
35. The n * n identity matrix is the transition matrix for a regu- (A) Discuss the behavior of the state matrices S1, S2, S3, c
lar Markov chain. for the initial-state matrix S0 = 3.2 .3 .54.
1 (B) Repeat part (A) for S0 = 3 13 1 1
4.
36. The n * n matrix in which each entry equals is the transi- 3 3
n
tion matrix for a regular Markov chain. (C) Discuss the behavior of P , k = 2, 3, 4, c.
k
37. If the 2 * 2 matrix P is the transition matrix for a regular (D) Which of the conclusions of Theorem 1 are not valid for
Markov chain, then, at most, one of the entries of P is equal to 0. this matrix? Why is this not a contradiction?
38. If the 3 * 3 matrix P is the transition matrix for a regular Markov 49. The transition matrix for a Markov chain is
chain, then, at most, two of the entries of P are equal to 0.
39. If a transition matrix P for a Markov chain has a stationary 1 0 0
matrix S, then P is regular. P = C .2 .2 .6 S
40. If P is the transition matrix for a Markov chain, then P has a 0 0 1
unique stationary matrix.
(A) Show that R = 31 0 04 and S = 30 0 14 are
In Problems 41–44, approximate the stationary matrix S for both stationary matrices for P. Explain why this does not
each transition matrix P by computing powers of the transition contradict Theorem 1A.
matrix P. Round matrix entries to four decimal places. (B) Find another stationary matrix for P. [Hint: Consider
T = aR + 11- a2S, where 0 6 a 6 1.]
.51 .49 .68 .32
41. P = c d 42. P = c d (C) How many different stationary matrices does P have?
.27 .73 .19 .81
50. The transition matrix for a Markov chain is 56. Transportation. The railroad in Problem 55 also has a fleet
.7 0 .3 of tank cars. If 14% of the tank cars on the home tracks enter
the national pool each month, and 26% of the tank cars in
P = C0 1 0S
the national pool are returned to the home tracks each month,
.2 0 .8 what percentage of its tank cars can the railroad expect to
have on its home tracks in the long run?
(A) Show that R = 3.4 0 .64 and S = 30 1 04 are
both stationary matrices for P. Explain why this does not 57. Labor force. Table 1 gives the percentage of the U.S. female
contradict Theorem 1A. population who were members of the civilian labor force
in the indicated years. The following transition matrix P is
(B) Find another stationary matrix for P. [Hint: Consider
proposed as a model for the data, where L represents females
T = aR + 11- a2S, where 0 6 a 6 1.]
who are in the labor force and L′ represents females who are
(C) How many different stationary matrices does P have? not in the labor force:
Problems 51 and 52 require the use of a graphing calculator. Next decade
L L′
51. Refer to the transition matrix P in Problem 49. What matrix
Current L .92 .08
P do the powers of P appear to be approaching? Are the rows c d = P
of P stationary matrices for P? decade L′ .2 .8
52. Refer to the transition matrix P in Problem 50. What matrix (A) Let S0 = 3.433 .5674, and find S1, S2, S3, and S4.
P do the powers of P appear to be approaching? Are the rows Compute the matrices exactly and then round entries to
of P stationary matrices for P? three decimal places.
53. The transition matrix for a Markov chain is (B) Construct a new table comparing the results from part
(A) with the data in Table 1.
.1 .5 .4
P = C .3 .2 .5 S (C) According to this transition matrix, what percentage of
the U.S. female population will be in the labor force in
.7 .1 .2
the long run?
Let Mk denote the maximum entry in the second column of
P k. Note that M1 = .5. Table 1
(A) Find M2, M3, M4, and M5 to three decimal places. Year Percent
1970 43.3
(B) Explain why Mk Ú Mk + 1 for all positive integers k.
1980 51.5
54. The transition matrix for a Markov chain is
1990 57.5
0 .2 .8 2000 59.8
P = C .3 .3 .4 S 2010 58.5
.6 .1 .3
58. Home ownership. The U.S. Census Bureau published the
Let mk denote the minimum entry in the third column of home ownership rates given in Table 2.
P k. Note that m1 = .3.
(A) Find m2, m3, m4, and m5 to three decimal places. Table 2
Year Percent
(B) Explain why mk … mk + 1 for all positive integers k.
1996 65.4
2000 67.4
2004 69.0
Applications 2008 67.8
55. Transportation. Most railroad cars are owned by individual The following transition matrix P is proposed as a model for the
railroad companies. When a car leaves its home railroad’s data, where H represents the households that own their home.
tracks, it becomes part of a national pool of cars and can be Four years later
used by other railroads. The rules governing the use of these
pooled cars are designed to eventually return the car to the H H′
home tracks. A particular railroad found that each month, 11% Current H .95 .05
c d = P
of its boxcars on the home tracks left to join the national pool, year H′ .15 .85
and 29% of its boxcars in the national pool were returned to
the home tracks. If these percentages remain valid for a long (A) Let S0 = 3.654 .3464 and find S1, S2, and S3. Compute
period of time, what percentage of its boxcars can this railroad both matrices exactly and then round entries to three
expect to have on its home tracks in the long run? decimal places.
(B) Construct a new table comparing the results from part in the stationary matrix correct to two decimal places. If
(A) with the data in Table 2. Acme decides to market this new soap, what is the long-run
expected total market share for their two soaps?
(C) According to this transition matrix, what percentage of
households will own their home in the long run? 64. Market share. Refer to Problem 63. The chemists at Acme
Soap Company have developed a second new soap, called
59. Market share. Consumers can choose between three long- brand Y. Test-marketing this soap against the three estab-
distance telephone services: GTT, NCJ, and Dash. Aggressive lished brands produces the following transition matrix:
marketing by all three companies results in a continual shift of
customers among the three services. Each year, GTT loses 5% SB DB SA Y
of its customers to NCJ and 20% to Dash, NCJ loses 15% of SB .3 .2 .2 .3
its customers to GTT and 10% to Dash, and Dash loses 5% of P = DB D .2 .2 .2 .4
T
its customers to GTT and 10% to NCJ. Assuming that these SA .2 .2 .4 .2
percentages remain valid over a long period of time, what is Y .1 .2 .3 .4
each company’s expected market share in the long run?
Proceed as in Problem 63 to approximate the elements in
60. Market share. Consumers in a certain area can choose be-
the stationary matrix correct to two decimal places. If Acme
tween three package delivery services: APS, GX, and WWP.
decides to market brand Y, what is the long-run expected total
Each week, APS loses 10% of its customers to GX and 20%
market share for Standard Acme and brand Y? Should Acme
to WWP, GX loses 15% of its customers to APS and 10% to
market brand X or brand Y?
WWP, and WWP loses 5% of its customers to APS and 5%
to GX. Assuming that these percentages remain valid over a 65. Genetics. A given plant species has red, pink, or white
long period of time, what is each company’s expected market flowers according to the genotypes RR, RW, and WW,
share in the long run? respectively. If each of these genotypes is crossed with a pink-
flowering plant (genotype RW), then the transition matrix is
61. Insurance. An auto insurance company classifies its cus-
tomers in three categories: poor, satisfactory, and preferred. Next generation
Each year, 40% of those in the poor category are moved to Red Pink White
satisfactory, and 20% of those in the satisfactory category are
This Red .5 .5 0
moved to preferred. Also, 20% in the preferred category are
moved to the satisfactory category, and 20% in the satisfac- generation Pink C .25 .5 .25 S
tory category are moved to the poor category. Customers are White 0 .5 .5
never moved from poor to preferred, or conversely, in a single
Assuming that the plants of each generation are crossed only
year. Assuming that these percentages remain valid over a
with pink plants to produce the next generation, show that
long period of time, how many customers are expected in
regardless of the makeup of the first generation, the genotype
each category in the long run?
composition will eventually stabilize at 25% red, 50% pink,
62. Insurance. Repeat Problems 61 if 40% of preferred custom- and 25% white. (Find the stationary matrix.)
ers are moved to the satisfactory category each year, and all 66. Gene mutation. Suppose that a gene in a chromosome is
other information remains the same. of type A or type B. Assume that the probability that a gene
Problems 63 and 64 require the use of a graphing calculator. of type A will mutate to type B in one generation is 10-4 and
that a gene of type B will mutate to type A is 10-6.
63. Market share. Acme Soap Company markets one brand of
soap, called Standard Acme (SA), and Best Soap Company (A) What is the transition matrix?
markets two brands, Standard Best (SB) and Deluxe Best (DB). (B) After many generations, what is the probability that the gene
Currently, Acme has 40% of the market, and the remainder is will be of type A? Of type B? (Find the stationary matrix.)
divided equally between the two Best brands. Acme is consid-
ering the introduction of a second brand to get a larger share 67. Rapid transit. A new rapid transit system has just started
of the market. A proposed new brand, called brand X, was operating. In the first month of operation, it is found that 25%
test-marketed in several large cities, producing the following of commuters are using the system, while 75% still travel
transition matrix for the consumers’ weekly buying habits: by car. The following transition matrix was determined from
records of other rapid transit systems:
SB DB SA X
Next month
SB .4 .1 .3 .2 Rapid
DB .3 .2 .2 .3 transit Car
P = D T
SA .1 .2 .2 .5 Current Rapid transit .8 .2
c d
X .3 .3 .1 .3 month Car .3 .7
Assuming that P represents the consumers’ buying habits (A) What is the initial-state matrix?
over a long period of time, use this transition matrix and the
(B) What percentage of commuters will be using the new
initial-state matrix S0 = 3.3 .3 .4 04 to compute suc-
system after 1 month? After 2 months?
cessive state matrices in order to approximate the elements
(C) Find the percentage of commuters using each type of Next decade
transportation after the new system has been in service S S′
for a long time. Current S .61 .39
c d = P
decade S′ .21 .79
68. Politics: filibuster. The Senate is in the middle of a floor
debate, and a filibuster is threatened. Senator Hanks, who is (A) Let S0 = 3.309 .6914 and find S1, S2, S3, and S4.
still vacillating, has a probability of .1 of changing his mind Compute the matrices exactly and then round entries to
during the next 5 minutes. If this pattern continues for each three decimal places.
5 minutes that the debate continues and if a 24-hour filibuster
takes place before a vote is taken, what is the probability that (B) Construct a new table comparing the results from part
Senator Hanks will cast a yes vote? A no vote? (A) with the data in Table 3.
(A) Complete the following transition matrix: (C) According to this transition matrix, what percentage of
the population will live in the south region in the long
Next 5 minutes
run?
Yes No
Current Yes .9 .1 70. Population shifts. Table 4 gives the percentage of the U.S.
c d population living in the northeast region during the indicated
5 minutes No
years.
(B) Find the stationary matrix and answer the two questions.
Table 4
(C) What is the stationary matrix if the probability of
Year Percent
Senator Hanks changing his mind (.1) is replaced with
an arbitrary probability p? 1970 24.1
1980 21.7
The population center of the 48 contiguous states of the United
1990 20.4
States is the point where a flat, rigid map of the contiguous states
would balance if the location of each person was represented on 2000 19.0
the map by a weight of equal measure. In 1790, the population 2010 17.9
center was 23 miles east of Baltimore, Maryland. By 1990, the
The following transition matrix P is proposed as a model for
center had shifted about 800 miles west and 100 miles south to a
the data, where N represents the population that lives in the
point in southeast Missouri. To study this shifting population, the
northeast region:
U.S. Census Bureau divides the states into four regions as shown
Next decade
in the figure. Problems 69 and 70 deal with population shifts
among these regions. N N′
Current N .61 .39
c d = P
West Midwest decade N′ .09 .91
Northeast
(A) Let S0 = 3.241 .7594 and find S1, S2, S3, and S4.
Compute the matrices exactly and then round entries to
three decimal places.
(B) Construct a new table comparing the results from part
(C) According to this transition matrix, what percentage of
South the population will live in the northeast region in the
Figure for 69 and 70.: Regions of the United long run?
States and the center of population
69. Population shifts. Table 3 gives the percentage of the U.S. Answers to Matched Problems
population living in the south region during the indicated years. 1. (A) Regular (B) Not regular (C) Regular
.2 .8
Table 3 2. S = 3.2 .84; P = c d
.2 .8
Year Percent
1970 30.9
3. 97.5%
1980 33.3 4. 28% preferred, 43% standard, 29% infrequent
1990 34.4 .1618 .2941 .5441
2000 35.6 5. P = C .1618 .2941 .5441 S ;
2010 37.1 .1618 .2941 .5441
The following transition matrix P is proposed as a model for S = 3.1618 .2941 .54414
the data, where S represents the population that lives in the
south region:
SECTION 8.3 Absorbing Markov Chains 459
8.3 Absorbing Markov Chains

■■ Absorbing States and Absorbing In Section 8.2, we saw that the powers of a regular transition matrix always approach
Chains a limiting matrix. Not all transition matrices have this property. In this section, we
■■ Standard Form discuss another type of Markov chain, called an absorbing Markov chain. Although
regular and absorbing Markov chains have some differences, they have one important
■■ Limiting Matrix similarity: the powers of the transition matrix for an absorbing Markov chain also
■■ Graphing Calculator approach a limiting matrix. After introducing basic concepts, we develop methods
Approximations for finding the limiting matrix and discuss the relationship between the states in the
Markov chain and the entries in the limiting matrix.
Absorbing States and Absorbing Chains

A state in a Markov chain is called an absorbing state if, once the state is entered, it
is impossible to leave.
EXAMPLE 1 Recognizing Absorbing States Identify any absorbing states for the following
transition matrices:
(A) A B C (B) A B C
A 1 0 0 A 0 0 1
P = B C .5 .5 0S P = B C0 1 0S
C 0 .5 .5 C 1 0 0
SOLUTION
(A) The probability of going from state A to state A is 1, and the probability of going
from state A to either state B or state C is 0. Once state A is entered, it is impos-
sible to leave, so A is an absorbing state. Since the probability of going from
state B to state A is nonzero, it is possible to leave B, and B is not an absorbing
state. Similarly, the probability of going from state C to state B is nonzero, so C
is not an absorbing state.
(B) Reasoning as before, the 1 in row 2 and column 2 indicates that state B is an
absorbing state. The probability of going from state A to state C and the prob-
ability of going from state C to state A are both nonzero. So A and C are not
absorbing states.
Matched Problem 1 Identify any absorbing states for the following transition
matrices:
(A) A B C (B) A B C
A .5 0 .5 A 0 1 0
P = BC0 1 0S P = B C1 0 0S
C 0 .5 .5 C 0 0 1
The reasoning used to identify absorbing states in Example 1 is generalized in

Theorem 1.
THEOREM 1 Absorbing States and Transition Matrices

A state in a Markov chain is absorbing if and only if the row of the transi-
tion matrix corresponding to the state has a 1 on the main diagonal and 0’s
elsewhere.
The presence of an absorbing state in a transition matrix does not guarantee that
the powers of the matrix approach a limiting matrix nor that the state matrices in the
corresponding Markov chain approach a stationary matrix. For example, if we square
the matrix P from Example 1B, we obtain
0 0 1 0 0 1 1 0 0
2
P = C0 1 0S C0 1 0S = C0 1 0S = I
1 0 0 1 0 0 0 0 1
Since P 2 = I, the 3 * 3 identity matrix, it follows that

P 3 = PP 2 = PI = P Since P 2 = I
P 4 = PP 3 = PP = I Since P 3 = P and PP = P 2 = I
In general, the powers of this transition matrix P oscillate between P and I and do not
approach a limiting matrix.

(A) For the initial-state matrix S0 = 3a b c4, find the first four state matrices,
S1, S2, S3, and S4, in the Markov chain with transition matrix
0 0 1
P = C0 1 0S
1 0 0
(B) Do the state matrices appear to be approaching a stationary matrix? Discuss.
To ensure that transition matrices for Markov chains with one or more absorbing
states have limiting matrices, it is necessary to require the chain to satisfy one addi-
tional condition, as stated in the following definition.
DEFINITION Absorbing Markov Chains

A Markov chain is an absorbing chain if
1. There is at least one absorbing state; and
2. It is possible to go from each nonabsorbing state to at least one absorbing state
in a finite number of steps.
As we saw earlier, absorbing states are identified easily by examining the rows of
a transition matrix. It is also possible to use a transition matrix to determine whether
a Markov chain is an absorbing chain, but this can be a difficult task, especially if the
matrix is large. A transition diagram is often a more appropriate tool for determining
whether a Markov chain is absorbing. The next example illustrates this approach for
the two matrices discussed in Example 1.
EXAMPLE 2 Recognizing Absorbing Markov Chains Use a transition diagram to determine

whether P is the transition matrix for an absorbing Markov chain.
(A) A B C (B) A B C
A 1 0 0 A 0 0 1
P = B C .5 .5 0 S P = B C0 1 0S
C 0 .5 .5 C 1 0 0
SOLUTION
(A) From Example 1A, we know that A is the only absorbing state. The second con-
.5 B .5 dition in the definition of an absorbing Markov chain is satisfied if we can show
that it is possible to go from the nonabsorbing states B and C to the absorbing
1 A .5
state A in a finite number of steps. This is easily determined by drawing a tran-
sition diagram (Fig. 1). Examining the diagram, we see that it is possible to go
from state B to the absorbing state A in one step and from state C to the absorbing
C .5 state A in two steps. So P is the transition matrix for an absorbing Markov chain.
(B) Again, we draw the transition diagram (Fig. 2). From this diagram it is clear
Figure 1
that it is impossible to go from either state A or state C to the absorbing state B.
So P is not the transition matrix for an absorbing Markov chain.
B 1
A 1
C
1
Figure 2
Matched Problem 2 Use a transition diagram to determine whether P is the

transition matrix for an absorbing Markov chain.
(A) A B C (B) A B C
A .5 0 .5 A 0 1 0
P = B C0 1 0S P = B C1 0 0S
C 0 .5 .5 C 0 0 1

Determine whether each statement is true or false. Use examples and verbal argu-
ments to support your conclusions.
(A) A Markov chain with two states, one nonabsorbing and one absorbing, is always
an absorbing chain.
(B) A Markov chain with two states, both of which are absorbing, is always an
absorbing chain.
(C) A Markov chain with three states, one nonabsorbing and two absorbing, is
always an absorbing chain.
.6
.1 A B 1
Standard Form
.3 .5 The transition matrix for a Markov chain is not unique. Consider the transition dia-
gram in Figure 3. Since there are 4! = 24 different ways to arrange the four states in
this diagram, there are 24 different ways to write a transition matrix. (Some of these
.2 C D 1 matrices may have identical entries, but all are different when the row and column
.3 labels are taken into account.) For example, the following matrices M, N, and P are
Figure 3 three different transition matrices for this diagram.
A B C D D B C A B D A C
A .1 .6 .3 0 D 1 0 0 0 B 1 0 0 0
B 0 1 0 0 B 0 1 0 0 D 0 1 0 0
M = D T N = D T P = D T (1)
C .5 0 .2 .3 C .3 0 .2 .5 A .6 0 .1 .3
D 0 0 0 1 A 0 .6 .3 .1 C 0 .3 .5 .2
In matrices N and P, notice that all the absorbing states precede all the
nonabsorbing states. A transition matrix written in this form is said to be in standard
form. We will find standard forms very useful in determining limiting matrices for
absorbing Markov chains.
Reminder
DEFINITION Standard Forms for Absorbing Markov Chains
The identity matrix I in the defini-
tion is a square matrix: It has 1’s A transition matrix for an absorbing Markov chain is in standard form if the
on the principal diagonal and 0’s rows and columns are labeled so that all the absorbing states precede all the non-
elsewhere (the number of rows in I absorbing states. (There may be more than one standard form.) Any standard
is the number of absorbing states). form can always be partitioned into four submatrices:
The zero matrix 0 is not necessar- A N
ily square (the number of rows in 0 A I 0 A = all absorbing states
is the number of absorbing states, c d
and the number of columns is the N R Q N = all nonabsorbing states
number of nonabsorbing states).
where I is an identity matrix and 0 is a zero matrix.
Referring to the matrix P in (1), we see that the submatrices in this standard form are
B D A C
1 0 0 0
I = c d 0 = c d B 1 0 0 0
0 1 0 0
D 0 1 0 0
.6 0 .1 .3 P = D T
R = c d Q = c d A .6 0 .1 .3
0 .3 .5 .2 C 0 .3 .5 .2
Limiting Matrix
We will now discuss the long-run behavior of absorbing Markov chains.
EXAMPLE 3 Real Estate Development Two competing real estate companies are trying to buy
all the farms in a particular area for future housing development. Each year, 20% of
the farmers decide to sell to company A, 30% decide to sell to company B, and the
rest continue to farm their land. Neither company ever sells any of the farms they
purchase.
(A) Draw a transition diagram and determine whether or not the Markov chain is
absorbing.
(B) Write a transition matrix that is in standard form.
(C) If neither company owns any farms at the beginning of this competitive buying
process, estimate the percentage of farms that each company will purchase in
the long run.
(D) If company A buys 50% of the farms before company B enters the competitive
buying process, estimate the percentage of farms that each company will pur-
chase in the long run.
SOLUTION
(A) .5 A = sells to company A
B = sells to company B
C = continues farming
C
.2 .3
1 A B 1
The associated Markov chain is absorbing since there are two absorbing states, A
and B. It is possible to go from the nonabsorbing state C to either A or B in one step.
(B) We use the transition diagram to write a transition matrix that is in standard form:
A B C
A 1 0 0
P = B C 0 1 0 S Standard form
C .2 .3 .5
(C) At the beginning of the competitive buying process all the farmers are in state
C (own a farm). Thus, S0 = 30 0 14. The successive state matrices are
(multiplication details omitted):
S1 = S0P = 3.2 .3 .54
S2 = S1P = 3.3 .45 .254
S3 = S2P = 3.35 .525 .1254
S4 = S3P = 3.375 .5625 .06254
S5 = S4P = 3.3875 .58125 .031254
S6 = S5P = 3.39375 .590625 .0156254
S7 = S6P = 3.396875 .5953125 .00781254
S8 = S7P = 3.3984375 .59765625 .003906254
S9 = S8P = 3.39921875 .598828125 .0019531254
It appears that these state matrices are approaching the matrix
A B C
S = 3.4 .6 04
This indicates that in the long run, company A will acquire approximately 40%
of the farms and company B will acquire the remaining 60%.
(D) This time, at the beginning of the competitive buying process 50% of farmers
are already in state A and the rest are in state C. So S0 = 3.5 0 .54. The
successive state matrices are (multiplication details omitted):
S1 = S0P = 3.6 .15 .254
S2 = S1P = 3.65 .225 .1254
S3 = S2P = 3.675 .2625 .06254
S4 = S3P = 3.6875 .28125 .031254
S5 = S4P = 3.69375 .290625 .0156254
S6 = S5P = 3.696875 .2953125 .00781254
S7 = S6P = 3.6984375 .29765625 .003906254
S8 = S7P = 3.69921875 .298828125 .0019531254
These state matrices approach a matrix different from the one in part (C):
A B C
S′ = 3.7 .3 04
Because of its head start, company A will now acquire approximately 70% of
the farms and company B will acquire the remaining 30%.
Matched Problem 3 Repeat Example 3 if 10% of farmers sell to company A

each year, 40% sell to company B, and the remainder continue farming.
Recall from Theorem 1, Section 8.2, that the successive state matrices of a
regular Markov chain always approach a stationary matrix. Furthermore, this station-
ary matrix is unique. That is, changing the initial-state matrix does not change the
stationary matrix. The successive state matrices for an absorbing Markov chain also
approach a stationary matrix, but this matrix is not unique. To confirm this, consider
the transition matrix P and the state matrices S and S′ from Example 3:
A B C
A 1 0 0 A B C A B C
P = BC0 1 0S S = 3.4 .6 04 S′ = 3.7 .3 04
C .2 .3 .5
It turns out that S and S′ are both stationary matrices, as the following multiplications
verify:
1 0 0
SP = 3.4 .6 04 C 0 1 0 S = 3.4 .6 04 = S
.2 .3 .5
1 0 0
S′P = 3.7 .3 04 C 0 1 0 S = 3.7 .3 04 = S′
.2 .3 .5
In fact, this absorbing Markov chain has an infinite number of stationary matrices
(see Problems 57 and 58, Exercises 8.3).
Changing the initial-state matrix for an absorbing Markov chain can cause the
successive state matrices to approach a different stationary matrix.
In Section 8.2, we used the unique stationary matrix for a regular Markov chain
to find the limiting matrix P. Since an absorbing Markov chain can have many differ-
ent stationary matrices, we cannot expect this approach to work for absorbing chains.
However, it turns out that transition matrices for absorbing chains do have limiting
matrices, and they are not very difficult to find. Theorem 2 gives us the necessary
tools. The proof of this theorem is left for more advanced courses.
THEOREM 2 Limiting Matrices for Absorbing Markov Chains

If a standard form P for an absorbing Markov chain is partitioned as
I 0
P = c d
R Q
then P k approaches a limiting matrix P as k increases, where

I 0
P = c d
FR 0
The matrix F is given by F = 1I - Q2 -1 and is called the fundamental

matrix for P.
The identity matrix used to form the fundamental matrix F must be the same
size as the matrix Q.
EXAMPLE 4 Finding the Limiting Matrix

(A) Find the limiting matrix P for the standard form P found in Example 3.
(B) Use P to find the limit of the successive state matrices for S0 = 30 0 14.
(C) Use P to find the limit of the successive state matrices for S0 = 3.5 0 .54.
SOLUTION
(A) From Example 3, we have
1 0 0
I 0
P = C0 1 0S c d
R Q
.2 .3 .5
where
1 0 0
I = c d 0 = c d R = 3.2 .34 Q = 3.54
0 1 0
If I = 314 is the 1 * 1 identity matrix, then I - Q is also a 1 * 1 matrix;

F = 1I - Q2 -1 is simply the multiplicative inverse of the single entry in
I - Q. So
F = 1314 - 3.542 -1 = 3.54 -1 = 324
FR = 3243.2 .34 = 3.4 .64
and the limiting matrix is
A B C
A 1 0 0
I 0
P = B C0 1 0S c d
FR 0
C .4 .6 0
(B) Since the successive state matrices are given by Sk = S0P k (Theorem 1,
Section 8.1) and P k approaches P, it follows that Sk approaches
1 0 0
S0 P = 30 0 14 C 0 1 0 S = 3.4 .6 04
.4 .6 0
which agrees with the results in part (C) of Example 3.

(C) This time, the successive state matrices approach
1 0 0
S0 P = 3.5 0 .54 C 0 1 0 S = 3.7 .3 04
.4 .6 0
which agrees with the results in part (D) of Example 3.
Matched Problem 4 Repeat Example 4 for the standard form P found in

Matched Problem 3.
Recall that the limiting matrix for a regular Markov chain contains the long-run
probabilities of going from any state to any other state. This is also true for the
limiting matrix of an absorbing Markov chain. Let’s compare the transition matrix P
and its limiting matrix P from Example 4:
A B C A B C
A 1 0 0 A 1 0 0
P = B C0 1 0S approaches P = B C0 1 0S
C .2 .3 .5 C .4 .6 0
The rows of P and P corresponding to the absorbing states A and B are identical. That
is, if the probability of going from state A to state A is 1 at the beginning of the chain,
then this probability will remain 1 for all trials in the chain and for the limiting ma-
trix. The entries in the third row of P give the long-run probabilities of going from the
nonabsorbing state C to states A, B, or C.
The fundamental matrix F provides some additional information about an ab-
sorbing chain. Recall from Example 4 that F = 324. It can be shown that the entries
in F determine the average number of trials that it takes to go from a given nonab-
sorbing state to an absorbing state. In the case of Example 4, the single entry 2 in F
indicates that it will take an average of 2 years for a farmer to go from state C (owns
a farm) to one of the absorbing states (sells the farm). Some will reach an absorbing
state in 1 year, and some will take more than 2 years. But the average will be 2 years.
These observations are summarized in Theorem 3, which we state without proof.
THEOREM 3 Properties of the Limiting Matrix P

If P is a transition matrix in standard form for an absorbing Markov chain, F is the
fundamental matrix, and P is the limiting matrix, then
(A) The entry in row i and column j of P is the long-run probability of going from
state i to state j. For the nonabsorbing states, these probabilities are also the
entries in the matrix FR used to form P.
(B) The sum of the entries in each row of the fundamental matrix F is the aver-
age number of trials it will take to go from each nonabsorbing state to some
absorbing state.
(Note that the rows of both F and FR correspond to the nonabsorbing states in the
order given in the standard form P).
CONCEPTUAL INSIGHT
1. The zero matrix in the lower right corner of the limiting matrix P in Theorem 2
indicates that the long-run probability of going from any nonabsorbing
state to any other nonabsorbing state is always 0. That is, in the long run, all
elements in an absorbing Markov chain end up in one of the absorbing states.
2. If the transition matrix for an absorbing Markov chain is not in standard
form, it is still possible to find a limiting matrix (see Problems 53 and 54,
Exercises 8.3). However, it is customary to use standard form when inves-
tigating the limiting behavior of an absorbing chain.
Now that we have developed the necessary tools for analyzing the long-run be-
havior of an absorbing Markov chain, we apply these tools to an earlier application
(see Example 4, Section 8.1).
EXAMPLE 5 Student Retention The following transition diagram is for part-time students
enrolled in a university MBA program:
.1 E = entry-level students
.6 E D 1 A = advanced-level students
D = dropouts
.3 G = graduates
.1
.5 A G 1
.4
(A) In the long run, what percentage of entry-level students will graduate? What
percentage of advanced-level students will not graduate?
(B) What is the average number of years that an entry-level student will remain in
this program? An advanced-level student?
SOLUTION
(A) First, notice that this is an absorbing Markov chain with two absorbing states,
state D and state G. A standard form for this absorbing chain is
D G E A
D 1 0 0 0
G 0 1 0 0 I 0
P = D T c d
E .1 0 .6 .3 R Q
A .1 .4 0 .5
The submatrices in this partition are

1 0 0 0 .1 0 .6 .3
I = c d 0 = c d R = c d Q = c d
0 1 0 0 .1 .4 0 .5
Therefore,
1 0 .6 .3 -1
F = 1I - Q2 -1 = a c d - c db
0 1 0 .5
.4 - .3 -1 Use row
= c d operations to
0 .5
find this matrix
2.5 1.5 inverse.
= c d
0 2
and
2.5 1.5 .1 0 .4 .6
FR = c dc d = c d
0 2 .1 .4 .2 .8
The limiting matrix is

D G E A
D 1 0 0 0
G 0 1 0 0 I 0
P = D T c d
E .4 .6 0 0 FR 0
A .2 .8 0 0
From this limiting form, we see that in the long run 60% of the entry-level students
will graduate and 20% of the advanced-level students will not graduate.
(B) The sum of the first-row entries of the fundamental matrix F is 2.5 + 1.5 = 4.
According to Theorem 3, this indicates that an entry-level student will spend
an average of 4 years in the transient states E and A before reaching one of the
absorbing states, D or G. The sum of the second-row entries of F is 0 + 2 = 2.
So an advanced-level student spends an average of 2 years in the program
before either graduating or dropping out.
Matched Problem 5 Repeat Example 5 for the following transition diagram:

.2
.5 E D 1 E = entry-level students
A = advanced-level students
.3 D = dropouts
.2 G = graduates
.6 A G 1
.2
Graphing Calculator Approximations

Just as was the case for regular Markov chains, the limiting matrix P for an absorbing
Markov chain with transition matrix P can be approximated by computing P k on a
graphing calculator for sufficiently large values of k. For example, computing P 50 for
the standard form P in Example 5 produces the following results:
1 0 0 0 50 1 0 0 0
0 1 0 0 0 1 0 0
P 50 = D T = D T = P
.1 0 .6 .3 .4 .6 0 0
.1 .4 0 .5 .2 .8 0 0
where we have replaced very small numbers displayed in scientific notation with 0
(see the Conceptual Insight on page 454, Section 8.2).
Before you use P k to approximate P, be certain to determine that
! CAUTION
P exists. If you attempt to approximate a limiting matrix when
none exists, the results can be misleading. For example, consider the transition matrix
1 0 0 0 0
.2 .2 0 .3 .3
P = E 0 0 0 .5 .5 U
0 0 1 0 0
0 0 1 0 0
Computing P 50 on a graphing calculator produces the following matrix:
1 0 0 0 0
.25 0 .625 .0625 .0625
P 50 = E0 0 1 0 0 U (2)
0 0 0 .5 .5
0 0 0 .5 .5
It is tempting to stop at this point and conclude that the matrix in (2) must be a good
approximation for P. But to do so would be incorrect! If P 50 approximates a limiting
matrix P, then P 51 should also approximate the same matrix. However, computing P 51
produces quite a different matrix:
1 0 0 0 0
.25 0 .125 .3125 .3125
P 51 = E0 0 0 .5 .5 U (3)
0 0 1 0 0
0 0 1 0 0
Computing additional powers of P shows that the even powers of P approach matrix
(2) while the odd powers approach matrix (3). The transition matrix P does not have a
limiting matrix.
A graphing calculator can be used to perform the matrix calculations necessary
to find P exactly, as illustrated in Figure 4 for the transition matrix P from Example 5.
The matrix names [I], [B], [C], and [F] denote I, R, Q, and F, respectively. This
approach has the advantage of producing the fundamental matrix F, whose row sums
provide additional information about the long-run behavior of the chain.
(A) Store I and R in the (B) Store Q in the (C) Compute F and FR
graphing calculator memory graphing calculator memory
Figure 4 Matrix calculations
Exercises 8.3
A In Problems 1–6, identify the absorbing states in the indicated 8. .5
transition matrix.
.5 A B .8
A B C A B C
A .6 .3 .1 A 0 1 0
.1 .2
1. P = B C 0 1 0S 2. P = B C .3 .2 .5 S C
C 0 0 1 C 0 0 1
A B C A B C
.9
A 0 0 1 A 1 0 0
3. P = B C 1 0 0S 4. P = B C .3 .4 .3 S 9. .4
C 0 1 0 C 0 0 1 .3 A B .1
A B C D A B C D
A 1 0 0 0 A 0 1 0 0 .7 .8 .5
B 0 0 1 0 B 1 0 0 0
5. P = D T 6. P = D T
C .1 .1 .5 .3 C .1 .2 .3 .4 .2 D C 1
D 0 0 0 1 D .7 .1 .1 .1
In Problems 7–10, identify the absorbing states for each transition
10. .4
diagram, and determine whether or not the diagram represents an
absorbing Markov chain. .3 A B .6
7. .7
.3 A B 1 .5
.2
.6 C 1 D C 1
.4
In Problems 11–20, could the given matrix be the transition ma- In Problems 25–28, find a standard form for the absorbing Mar-
trix of an absorbing Markov chain? kov chain with the indicated transition matrix.
A B C A B C
0 1 1 0 A .2 .3 .5 A 0 0 1
11. J R 12. J R
1 0 0 1 25. P = B C 1 0 0 S 26. P = B C 0 1 0 S
.3 .7 .6 .4 C 0 0 1 C .7 .2 .1
13. J R 14. J R
0 1 1 0 A B C D A B C D
A .1 .2 .3 .4 A 0 .3 .3 .4
1 0 0 0 1 0
B 0 1 0 0 B 0 1 0 0
15. C 0 1 0S 16. C 0 0 1S 27. P = D T 28. P = D T
C .5 .2 .2 .1 C 0 0 1 0
0 0 1 1 0 0
D 0 0 0 1 D .8 .1 .1 0
.9 .1 0 .5 .5 0 In Problems 29–34, find the limiting matrix for the indicated
17. C .1 .9 0S 18. C .4 .3 .3 S standard form. Find the long-run probability of going from each
0 0 1 0 0 1 nonabsorbing state to each absorbing state and the average
number of trials needed to go from each nonabsorbing state to an
.9 0 .1 1 0 0 absorbing state.
19. C 0 1 0S 20. C 0 0 1S
A B C A B C
0 .2 .8 0 .7 .3
A 1 0 0 A 1 0 0
29. P = B C 0 1 0S 30. P = B C 0 1 0S
B In Problems 21–24, find a standard form for the absorbing Mar- C .3
C .1 .4 .5 .2 .5
kov chain with the indicated transition diagram.
A B C A B C
21. .5
A 1 0 0 A 1 0 0
.2 A B 1
.3 31. P = B C .2 .6 .2 S 32. P = B C .1 .6 .3 S
C .4 .2 .4 C .2 .2 .6
.5 C .1
A B C D A B C D
A 1 0 0 0 A 1 0 0 0
.4 B 0 1 0 0 B 0 1 0 0
33. P = D T 34. P = D T
22. .1 C .1 .2 .6 .1 C .1 .1 .7 .1
.2 A B 1 D .2 .2 .3 .3 D .3 .1 .4 .2
Problems 35–40 refer to the matrices in Problems 29–34. Use the

.7 C limiting matrix P found for each transition matrix P in Problems
29–34 to determine the long-run behavior of the successive state
matrices for the indicated initial-state matrices.
1
35. For matrix P from Problem 29 with
23. .4
(A) S0 = 30 0 14 (B) S0 = 3.2 .5 .34
.3 A B 1
(A) S0 = 30 0 14 (B) S0 = 3.2 .5 .34
.1 .4
.2 37. For matrix P from Problem 31 with
(A) S0 = 30 0 14 (B) S0 = 3.2 .5 .34
1 D C .3 38. For matrix P from Problem 32 with
.3 (A) S0 = 30 0 14 (B) S0 = 3.2 .5 .34
24. .4
(A) S0 = 30 0 0 14 (B) S0 = 30 0 1 04
.6 A B 1
(C) S0 = 30 0 .4 .64 (D) S0 = 3.1 .2 .3 .44
.7
(A) S0 = 30 0 0 14 (B) S0 = 30 0 1 04
1
D C
(C) S0 = 30 0 .4 .64 (D) S0 = 3.1 .2 .3 .44
.3
In Problems 41–48, discuss the validity of each statement. If the Step 2 Find the limiting matrix for this standard form.
statement is always true, explain why. If not, give a counterexample. Step 3 Using a transition diagram, reverse the process used in
41. If a Markov chain has an absorbing state, then it is an absorbing Step 1 to produce a limiting matrix for the original matrix P.
chain. 54. Repeat Problem 53 for
42. If a Markov chain has exactly two states and at least one A B C D
absorbing state, then it is an absorbing chain. A 1 0 0 0
43. If a Markov chain has exactly three states, one absorbing and B .3 .6 0 .1
P = D T
two nonabsorbing, then it is an absorbing chain. C .2 .3 .5 0
44. If a Markov chain has exactly three states, one nonabsorbing D 0 0 0 1
and two absorbing, then it is an absorbing chain.
55. Verify the results in Problem 53 by computing P k on a graph-
45. If every state of a Markov chain is an absorbing state, then it ing calculator for large values of k.
is an absorbing chain.
56. Verify the results in Problem 54 by computing P k on a graph-
46. If a Markov chain is absorbing, then it has a unique station- ing calculator for large values of k.
ary matrix.
57. Show that S = 3x 1 - x 04, 0 … x … 1, is a stationary
47. If a Markov chain is absorbing, then it is regular. matrix for the transition matrix
48. If a Markov chain is regular, then it is absorbing. A B C
A 1 0 0
In Problems 49–52, use a graphing calculator to approximate the P = B C0 1 0S
limiting matrix for the indicated standard form.
C .1 .5 .4
A B C D A B C D
Discuss the generalization of this result to any absorbing
A 1 0 0 0 A 1 0 0 0
Markov chain with two absorbing states and one nonabsorb-
B 0 1 0 0 B 0 1 0 0 ing state.
49. P = D T 50. P = D T
C .5 .3 .1 .1 C .1 .1 .5 .3
58. Show that S = 3x 1 - x 0 04, 0 … x … 1, is a sta-
D .6 .2 .1 .1 D 0 .2 .3 .5
tionary matrix for the transition matrix
A B C D E A B C D
A 1 0 0 0 0 A 1 0 0 0
B 0 1 0 0 0 B 0 1 0 0
P = D T
51. P = C E 0 .4 .5 0 .1 U C .1 .2 .3 .4
D 0 .4 0 .3 .3 D .6 .2 .1 .1
E .4 .4 0 .2 0 Discuss the generalization of this result to any absorbing
Markov chain with two absorbing states and two nonabsorb-
A B C D E
ing states.
A 1 0 0 0 0
B 0 1 0 0 0 59. An absorbing Markov chain has the following matrix P as a
standard form:
52. P = C E .5 0 0 0 .5 U
D 0 .4 0 .2 .4 A B C D
E 0 0 .1 .7 .2 A 1 0 0 0
B .2 .3 .1 .4 I 0
P = D T c d
C 53. The following matrix P is a nonstandard transition matrix for C 0 .5 .3 .2 R Q
an absorbing Markov chain: D 0 .1 .6 .3
A B C D
Let wk denote the maximum entry in Qk. Note that w1 = .6.
A .2 .2 .6 0
B 0 1 0 0 (A) Find w2, w4, w8, w16, and w32 to three decimal places.
P = D T
C .5 .1 0 .4 (B) Describe Qk when k is large.
D 0 0 0 1 60. Refer to the matrices P and Q of Problem 59. For k a positive
integer, let Tk = I + Q + Q2 + g + Qk.
To find a limiting matrix for P, follow the steps outlined
below. (A) Explain why Tk + 1 = TkQ + I.
Step 1 Using a transition diagram, rearrange the columns (B) Using a graphing calculator and part (A) to quickly com-
and rows of P to produce a standard form for this chain. pute the matrices Tk, discover and describe the connec-
tion between 1I - Q2 -1 and Tk when k is large.
(A) In the long run, what is the market share of each

Applications company?
61. Loans. A credit union classifies car loans into one of four (B) On average, how many years will it take a department to
categories: the loan has been paid in full (F), the account is in decide to use calculators from one of these companies in
good standing (G) with all payments up to date, the account their courses?
is in arrears (A) with one or more missing payments, or the 64. Pensions. Once a year company employees are given the op-
account has been classified as a bad debt (B) and sold to a portunity to join one of three pension plans: A, B, or C. Once
collection agency. Past records indicate that each month 10% an employee decides to join one of these plans, the employee
of the accounts in good standing pay the loan in full, 80% cannot drop the plan or switch to another plan. Past records
remain in good standing, and 10% become in arrears. Fur- indicate that each year 4% of employees elect to join plan
thermore, 10% of the accounts in arrears are paid in full, 40% A, 14% elect to join plan B, 7% elect to join plan C, and the
become accounts in good standing, 40% remain in arrears, remainder do not join any plan.
and 10% are classified as bad debts.
(A) In the long run, what percentage of the employees will
(A) In the long run, what percentage of the accounts in ar- elect to join plan A? Plan B? Plan C?
rears will pay their loan in full?
(B) On average, how many years will it take an employee to
(B) In the long run, what percentage of the accounts in good decide to join a plan?
standing will become bad debts?
65. Medicine. After bypass surgery, patients are placed in an
(C) What is the average number of months that an account in intensive care unit (ICU) until their condition stabilizes.
arrears will remain in this system before it is either paid Then they are transferred to a cardiac care ward (CCW),
in full or classified as a bad debt? where they remain until they are released from the hospital.
62. Employee training. A chain of car muffler and brake repair In a particular metropolitan area, a study of hospital records
shops maintains a training program for its mechanics. All produced the following data: each day 2% of the patients
new mechanics begin training in muffler repairs. Every in the ICU died, 52% were transferred to the CCW, and the
3 months, the performance of each mechanic is reviewed. remainder stayed in the ICU. Furthermore, each day 4% of
Past records indicate that after each quarterly review, 30% of the patients in the CCW developed complications and were
the muffler repair trainees are rated as qualified to repair muf- returned to the ICU, 1% died while in the CCW, 22% were
flers and begin training in brake repairs, 20% are terminated released from the hospital, and the remainder stayed in the
for unsatisfactory performance, and the remainder continue CCW.
as muffler repair trainees. Also, 30% of the brake repair train- (A) In the long run, what percentage of the patients in the
ees are rated as fully qualified mechanics requiring no further ICU are released from the hospital?
training, 10% are terminated for unsatisfactory performance,
and the remainder continue as brake repair trainees. (B) In the long run, what percentage of the patients in the
CCW die without ever being released from the hospital?
(A) In the long run, what percentage of muffler repair train-
ees will become fully qualified mechanics? (C) What is the average number of days that a patient in the
ICU will stay in the hospital?
(B) In the long run, what percentage of brake repair trainees
will be terminated? 66. Medicine. The study discussed in Problem 65 also produced
the following data for patients who underwent aortic valve
(C) What is the average number of quarters that a muffler replacements: each day 2% of the patients in the ICU died,
repair trainee will remain in the training program before 60% were transferred to the CCW, and the remainder stayed
being either terminated or promoted to fully qualified in the ICU. Furthermore, each day 5% of the patients in the
mechanic? CCW developed complications and were returned to the ICU,
63. Marketing. Three electronics firms are aggressively market- 1% died while in the CCW, 19% were released from the
ing their graphing calculators to high school and college hospital, and the remainder stayed in the CCW.
mathematics departments by offering volume discounts, com- (A) In the long run, what percentage of the patients in the
plimentary display equipment, and assistance with curricu- CCW are released from the hospital?
lum development. Due to the amount of equipment involved
and the necessary curriculum changes, once a department (B) In the long run, what percentage of the patients in the
decides to use a particular calculator in their courses, they ICU die without ever being released from the hospital?
never switch to another brand or stop using calculators. Each (C) What is the average number of days a patient in the
year, 6% of the departments decide to use calculators from CCW will stay in the hospital?
company A, 3% decide to use calculators from company
B, 11% decide to use calculators from company C, and the 67. Psychology. A rat is placed in room F or room B of the
remainder decide not to use any calculators in their courses. maze shown in the figure. The rat wanders from room to
room until it enters one of the rooms containing food, L or R.
Assume that the rat chooses an exit from a room at random 3. (A) .5
and that once it enters a room with food it never leaves.
(A) What is the long-run probability that a rat placed in
C
room B ends up in room R? .1 .4
(B) What is the average number of exits that a rat placed in

1 A B 1
room B will choose until it finds food?
(B) A B C
A 1 0 0
L P = BC0 1 0S
B C .1 .4 .5
R
(C) Company A will purchase 20% of the farms, and
F
company B will purchase 80%.
(D) Company A will purchase 60% of the farms, and
company B will purchase 40%.
4. (A) A B C
A 1 0 0
Figure for 67 and 68
P = B C0 1 0S
68. Psychology. Repeat Problem 67 if the exit from room B to C .2 .8 0
room R is blocked.
(B) [.2 .8 04
(C) [.6 .4 04
5. (A) Thirty percent of entry-level students will graduate;
1. (A) State B is absorbing.
50% of advanced-level students will not graduate.
(B) State C is absorbing.
(B) An entry-level student will spend an average of 3.5 years
2. (A) Absorbing Markov chain in the program; an advanced-level student will spend an
(B) Not an absorbing Markov chain average of 2.5 years in the program.

8.1 Properties of Markov Chains EXAMPLES
• The progression of a system through a sequence of states is called a stochastic process if chance ele-
ments are involved in the transition from one state to the next.
• A transition diagram or transition probability matrix can be used to represent the probabilities of
moving from one state to another. If those probabilities do not change with time, the stochastic process is
called a Markov chain.
• If a Markov chain has n states, then the entry ski of the kth-state matrix
Sk = 3sk1 sk2 g skn 4

gives the probability of being in state i after the kth trial. The sum of the entries in Sk is 1. Ex. 1, p. 440
• The entry pi, j of the n * n transition matrix P gives the probability of moving from state i to state j on Ex. 2, p. 441
the next trial. The sum of the entries in each row of P is 1. Ex. 3, p. 442
k
• If S0 is an initial-state matrix for a Markov chain, then Sk = S0P (Theorem 1, page 441). Ex. 4, p. 442

• A transition matrix P is regular if some power of P has only positive entries. Ex. 1, p. 449
• A Markov chain is a regular Markov chain if its transition matrix is regular.
• A state matrix S is stationary if SP = S. Ex. 2, p. 450
• The state matrices for a regular Markov chain approach a unique stationary matrix S (Theorem 1, page 450). Ex. 3, p. 451
k
• If P is the transition matrix for a regular Markov chain, then the matrices P approach a limiting matrix Ex. 4, p. 452
P, where each row of P is equal to the unique stationary matrix S (Theorem 1, page 450). Ex. 5, p. 453
8.3 Absorbing Markov Chains
• A state in a Markov chain is an absorbing state if once the state is entered it is impossible to leave. Ex. 1, p. 459
A state is absorbing if and only if its row in the transition matrix has a 1 on the main diagonal and 0’s
elsewhere.
• A Markov chain is an absorbing Markov chain if there is at least one absorbing state and it is possible Ex. 2, p. 460
to go from each nonabsorbing state to at least one absorbing state in a finite number of steps.
• A transition matrix for an absorbing Markov chain is in standard form if the rows and columns are
labeled so that all the absorbing states precede all the nonabsorbing states.
• If a standard form P for an absorbing Markov chain is partitioned as Ex. 3, p. 462
Ex. 4, p. 465
I 0
P = c d
R Q
then P k approaches a limiting matrix P as k increases, where
I 0
P = c d
FR 0
The matrix F = 1I - Q2 -1, where I is the identity matrix of the same size as Q, is called the Ex. 5, p. 467
fundamental matrix for P (Theorem 2, page 464).
• The entry in row i and column j of P is the long-run probability of going from state i to state j. The sum
of the entries in each row of F is the average number of trials that it will take to go from each nonabsorb-
ing state to some absorbing state (Theorem 3, page 466).
Review Exercises
Work through all the problems in this chapter review and check A B C
A B
your answers in the back of the book. Answers to all review prob- A .8 0 .2
A 0 1
lems are there along with section numbers in italics to indicate 4. P = c d 5. P = B C 0 1 0S
where each type of problem is discussed. Where weaknesses show B 1 0
C 0 0 1
A B C D
A 1. Given the transition matrix P and initial-state matrix S0 shown A 1 0 0 0
below, find S1 and S2 and explain what each represents:
B 0 1 0 0
6. P = D T
A B C 0 0 .3 .7
A .6 .4 D 0 0 .6 .4
P = c d S0 = [.3 .7]
B .2 .8
In Problems 7–10, write a transition matrix for the transition dia-
In Problems 2–6, P is a transition matrix for a Markov chain. gram indicated, identify any absorbing states, and classify each
Identify any absorbing states and classify the chain as regular, Markov chain as regular, absorbing, or neither.
absorbing, or neither. 7. 8.
A B A B B 1 B .2
1
A 1 0 A 0 1
2. P = c d 3. P = c d .1 .1
B .7 .3 B .7 .3
A .9 1 A .7
C C 1
9. In Problems 17–20, use a graphing calculator to approximate the

.1 B .2 limiting matrix for the indicated transition matrix.
17. Matrix P from Problem 13
A .7 1 18. Matrix P from Problem 14
19. Matrix P from Problem 15
1 C 20. Matrix P from Problem 16
21. Find a standard form for the absorbing Markov chain with
10. .2 transition matrix
.3 A B 1
A B C D
A.6 .1 .2 .1
.5 B 0 1 0 0
P = D T
.3 C .3 .2 .3 .2
.7 D C .2 D 0 0 0 1
.8 In Problems 22 and 23, determine the long-run behavior of the

successive state matrices for the indicated transition matrix and
11. A Markov chain has three states, A, B, and C. The probability initial-state matrices.
of going from state A to state B in one trial is .2, the probabil-
A B C
ity of going from state A to state C in one trial is .5, the prob-
A 0 1 0
ability of going from state B to state A in one trial is .8, the
probability of going from state B to state C in one trial is .2, 22. P = B C 0 0 1S
the probability of going from state C to state A in one trial is C .2 .6 .2
.1, and the probability of going from state C to state B in one (A) S0 = [0 0 1]
trial is .3. Draw a transition diagram and write a transition (B) S0 = [.5 .3 .2]
matrix for this chain.
A B C
12. Given the transition matrix A 1 0 0
A B 23. P = B C 0 1 0 S
A .4 .6 C .2 .6 .2
P = J R
B .9 .1 (A) S0 = [0 0 1]
find the probability of (B) S0 = [.5 .3 .2]
(A) Going from state A to state B in two trials 24. Let P be a 2 * 2 transition matrix for a Markov chain. Can P
(B) Going from state B to state A in three trials be regular if two of its entries are 0? Explain.
B In Problems 13 and 14, solve the equation SP = S to find the 25. Let P be a 3 * 3 transition matrix for a Markov chain. Can
stationary matrix S and the limiting matrix P. P be regular if three of its entries are 0? If four of its entries
A B C are 0? Explain.
A B A .4 .6 0
A .4 .6 C 26. A red urn contains 2 red marbles, 1 blue marble, and 1 green
13. P = J R 14. P = B C .5 .3 .2 S
B .2 .8 marble. A blue urn contains 1 red marble, 3 blue marbles, and
C 0 .8 .2
1 green marble. A green urn contains 6 red marbles, 3 blue
marbles, and 1 green marble. A marble is selected from an
In Problems 15 and 16, find the limiting matrix for the indicated urn, the color is noted, and the marble is returned to the urn
standard form. Find the long-run probability of going from each from which it was drawn. The next marble is drawn from the
nonabsorbing state to each absorbing state and the average urn whose color is the same as the marble just drawn. This
number of trials needed to go from each nonabsorbing state to an is a Markov process with three states: draw from the red urn,
absorbing state. draw from the blue urn, or draw from the green urn.
A B C D (A) Draw a transition diagram for this process.
A B C
A 1 0 0 0
A 1 0 0 (B) Write the transition matrix P.
B 0 1 0 0
15. P = B C 0 1 0 S 16. P = D T (C) Determine whether this chain is regular, absorbing, or
C .1 .5 .2 .2 neither.
C .3 .1 .6
D .1 .1 .4 .4 (D) Find the limiting matrix P, if it exists, and describe the
long-run behavior of this process.
27. Repeat Problem 26 if the blue and green marbles are removed (A) Draw a transition diagram.
from the red urn.
(A) Write a transition matrix.
28. Show that S = [x y z 0], where 0 … x … 1,
(B) Write the initial-state matrix.
0 … y … 1, 0 … z … 1, and x + y + z = 1, is a stationary
matrix for the transition matrix (C) Find the first-state matrix and explain what it represents.
A B C D (D) Find the stationary matrix.
A 1 0 0 0
(E) What percentage of the market will brand X have in the
B 0 1 0 0 long run if the transition matrix does not change?
P = D T
C 0 0 1 0
39. Marketing. Recent technological advances have led to
D .1 .3 .4 .2
the development of three new milling machines: brand A,
Discuss the generalization of this result to any absorbing brand B, and brand C. Due to the extensive retooling and
chain with three absorbing states and one nonabsorbing state. startup costs, once a company converts its machine shop
29. Give an example of a transition matrix for a Markov chain to one of these new machines, it never switches to another
that has no limiting matrix. brand. Each year, 6% of the machine shops convert to brand
A machines, 8% convert to brand B machines, 11% convert
30. Give an example of a transition matrix for an absorbing
to brand C machines, and the remainder continue to use their
Markov chain that has two different stationary matrices.
old machines.
31. Give an example of a transition matrix for a regular Markov
(A) In the long run, what is the market share of each brand?
chain for which 3.3 .1 .64 is a stationary matrix.
(B) What is the average number of years that a company
32. Give an example of a transition matrix for an absorbing
waits before converting to one of the new milling
Markov chain for which 3.3 .1 .64 is a stationary matrix.
machines?
33. Explain why an absorbing Markov chain that has more than
40. Internet. Table 1 gives the percentage of U.S. adults who at
one state is not regular.
least occasionally used the Internet in the given year.
34. Explain why a regular Markov chain that has more than one
state is not absorbing. Table 1
35. A Markov chain has transition matrix Year Percent
1995 14
.4 .6 2000 49
P = c d
.2 .8 2005 68
For S = 3.3 .94, calculate SP. Is S a stationary matrix? 2010 79
Explain. Source: Pew Internet &
American Life Project Surveys
In Problems 36 and 37, use a graphing calculator to approximate
the entries (to three decimal places) of the limiting matrix, if it The following transition matrix P is proposed as a model for
exists, of the indicated transition matrix. the data, where I represents the population of Internet users.
A B C D Five years later
A.2 .3 .1 .4 I I
B 0 0 1 0 Current I .95 .05
36. P = D T J R = P
C 0 .8 0 .2 year I′ .40 .60
D 0 0 1 0
(A) Let S0 = 3.14 .864 and find S1, S2, and S3. Compute
A B C D both matrices exactly and then round entries to two
A .1 0 .3 .6 decimal places.
B .2 .4 .1 .3
37. P = D T (B) Construct a new table comparing the results from part
C .3 .5 0 .2
D .9 .1 0 0
(C) According to this transition matrix, what percentage
of the adult U.S. population will use the Internet in the
Applications long run?
41. Employee training. In order to become a fellow of the
38. Product switching. A company’s brand (X) has 20% of the Society of Actuaries, a person must pass a series of ten
market. A market research firm finds that if a person uses examinations. Passage of the first two preliminary exams is
brand X, the probability is .7 that he or she will buy it next a prerequisite for employment as a trainee in the actuarial
time. On the other hand, if a person does not use brand X department of a large insurance company. Each year, 15%
(represented by X′), the probability is .5 that he or she will of the trainees complete the next three exams and become
switch to brand X next time. associates of the Society of Actuaries, 5% leave the com-
pany, never to return, and the remainder continue as trainees. 43. Smoking. Table 2 gives the percentage of U.S. adults who
Furthermore, each year, 17% of the associates complete the were smokers in the given year.
remaining five exams and become fellows of the Society of
Actuaries, 3% leave the company, never to return, and the Table 2
remainder continue as associates. Year Percent
(A) In the long run, what percentage of the trainees will 1985 30.1
become fellows? 1995 24.7
2005 20.9
(B) In the long run, what percentage of the associates will 2010 19.3
leave the company?
Source: American Lung
(C) What is the average number of years that a trainee re- Association
mains in this program before either becoming a fellow or The following transition matrix P is proposed as a model
being discharged? for the data, where S represents the population of U.S. adult
42. Genetics. A given plant species has red, pink, or white flow- smokers.
ers according to the genotypes RR, RW, and WW, respective- Five years later
ly. If each of these genotypes is crossed with a red-flowering S S′
plant, the transition matrix is
Current S .74 .26
J R = P
Next generation year S′ .03 .97
Red Pin White
Red 1 0 0 (A) Let S0 = 3.301 .6994, and find S1, S2, and S3.
This Compute the matrices exactly and then round entries to
Pink C .5 .5 0S
generation three decimal places.
White 0 1 0
(B) Construct a new table comparing the results from part
If each generation of the plant is crossed only with red plants
to produce the next generation, show that eventually all the (C) According to this transition matrix, what percentage
flowers produced by the plants will be red. (Find the limiting of the adult U.S. population will be smokers in the
matrix.) long run?
9 Data Description and
Probability Distributions
9.1 Graphing Data
Introduction
9.2 Measures of Central
In this chapter, we study various techniques for analyzing and displaying data. We
Tendency
use bar graphs, broken-line graphs, and pie graphs to present visual interpretations
9.3 Measures of Dispersion or comparisons of data. We use measures of central tendency (mean, median, and
9.4 Bernoulli Trials and mode) and measures of dispersion (range, variance, and standard deviation) to
Binomial Distributions describe and compare data sets.
Data collected from different sources—IQ scores, measurements of manu-
9.5 Normal Distributions
factured parts, healing times after surgery, for example—often exhibit surprising
similarity. We might express such similarity by saying that all three data sets
exhibit characteristics of a normal distribution. In Sections 9.4 and 9.5, we
develop theoretical probability distributions (binomial distributions and normal
distributions) that can be used as models of empirical data (see Problem 71 in
Section 9.5).
478
SECTION 9.1 Graphing Data 479
9.1 Graphing Data

■■ Bar Graphs, Broken-Line Graphs, Websites, television, newspapers, magazines, books, and reports make substantial use
and Pie Graphs of graphics to visually communicate complicated sets of data to the viewer. In this
■■ Frequency Distributions section, we look at bar graphs, broken-line graphs, and pie graphs and the techniques
for producing them. It is important to remember that graphs are visual aids and
■■ Comments on Statistics should be prepared with care. The object is to provide the viewer with the maximum
■■ Histograms amount of information while minimizing the time and effort required to interpret the
■■ Frequency Polygons and Cumulative
information in the graph.
Frequency Polygons
Bar Graphs, Broken-Line Graphs, and Pie Graphs
Table 1 U.S. Public Debt Bar graphs are used widely because they are easy to construct and interpret. They
are effective in presenting visual interpretations or comparisons of data. Consider
Year Debt (billion $)
Tables 1 and 2. Bar graphs are well suited to describe these two data sets. Vertical bars
1975 533
are usually used for time series—that is, data that changes over time, as in Table 1.
1985 1,823
The labels on the horizontal axis are then units of time (hours, days, years), as shown
1995 4,974
in Figure 1. Horizontal bars are generally used for data that changes by category, as
2005 7,934
2015 18,151
in Table 2, because of the ease of labeling categories on the vertical axis of the bar
graph (see Fig. 2). To increase clarity, space is left between the bars. Bar graphs for
Source: U.S. Treasury
the data in Tables 1 and 2 are illustrated in Figures 1 and 2, respectively.
Table 2 Passenger Boardings at Busiest U.S. Airports, 2015

Airport Boardings (million passengers)
Atlanta (ATL) 49.3
Chicago (ORD) 36.4
Los Angeles (LAX) 36.3
Dallas/Ft. Worth (DFW) 31.6
New York (JFK) 27.8
Source: Federal Aviation Administration
U.S. Public Debt
20,000 Passenger Boardings at Busiest U.S. Airports, 2015
Atlanta
15,000
Billion dollars
Chicago (O'Hare)
10,000
Los Angeles
5,000 Dallas/Fort Worth
New York
0
1975 1985 1995 2005 2015 0 10 20 30 40 50
Year Boardings (million passengers)
Figure 1 Vertical bar graph Figure 2 Horizontal bar graph
Two additional variations on bar graphs, the double bar graph and the divided bar
graph, are illustrated in Figures 3 and 4, respectively.
480 CHAPTER 9 Data Description and Probability Distributions
Education and Income, 2014

Population of World's Fastest Growing Megacities,
Bachelor's degree Male
Female 2016 and 2030 (projected)
Associate degree Dhaka, Bangladesh
Karachi, Pakistan
Some college
Lagos, Nigeria
High school diploma
Kinshasa, DR Congo
Some high school Bangalore, India
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 0 5 10 15 20 25 30 35
Median annual income ($) Million people
Figure 3 Double bar graph Figure 4 Divided bar graph
Source: Institute of Education Sciences Source: UN Population Division

(A) Using Figure 3, estimate the median annual income of a male with some college
and a female who holds a bachelor’s degree. Within which educational category
is there the greatest difference between male and female income? The least dif-
ference?
(B) Using Figure 4, estimate the population of Kinshasa in the years 2016 and 2030.
Which of the cities is projected to have the greatest increase in population from
2016 to 2030? The least increase? The greatest percentage increase? The least
percentage increase?
A broken-line graph can be obtained from a vertical bar graph by joining the
midpoints of the tops of consecutive bars with straight lines. For example, using
Figure 1, we obtain the broken-line graph in Figure 5.
U.S. Public Debt
20,000
15,000
Billion dollars
10,000
5,000
0
1975 1985 1995 2005 2015
Year
Figure 5 Broken-line graph
Broken-line graphs are particularly useful when we want to emphasize the change
in one or more variables relative to time. Figures 6 and 7 illustrate two additional
variations of broken-line graphs.
Revenue and Cost Projections of U.S. Energy Consumption

3 40 Petroleum
Revenue products
35
Profit Cost 30
Quadrillion BTU
Million dollars
2 Coal
25
Loss Natural
20 gas
15
1 Profit
10 Renewable
Nuclear
5
0 0
2010 2011 2012 2013 2014 2015 2016 2017 2018 2010 2015 2020 2025 2030
Year Year
Figure 6 Broken-line graphs (variation) Figure 7 Broken-line graphs (variation)

(A) Using Figure 6, estimate the revenue and costs in 2017. In which years is a
profit realized? In which year is the greatest loss experienced?
(B) Using Figure 7, estimate the U.S. consumption of nuclear energy in 2030. Esti-
mate the percentage of total consumption that will come from renewable energy
in the year 2030.
A pie graph is generally used to show how a whole is divided among several cat-
egories. The amount in each category is expressed as a percentage, and then a circle
is divided into segments (pieces of pie) proportional to the percentages of each cat-
egory. The central angle of a segment is the percentage of 360° corresponding to the
percentage of that category (see Fig. 8). In constructing pie graphs, we use relatively
few categories, usually arrange the segments in ascending or descending order of size
around the circle, and label each part.
Active U.S. Military Personnel, 2017 World Refugees, 2015

Middle East and
North Africa
17%
Navy Africa
Army 25% (908) 27%
36% (1308)
Europe
Air Force 27% Asia
25% (908) and Pacific
24%
Marine Corps Americas

14% (508) 5%
(A) (B)
Figure 8 Pie graphs
Bar graphs, broken-line graphs, and pie graphs are easily constructed using a
spreadsheet. After entering the data (see Fig. 9 for the data in Fig. 8A) and choosing
the type of display (bar, broken-line, pie), the graph is drawn automatically. Various
options for axes, gridlines, patterns, and text are available to improve the clarity of
the visual display.
Figure 9
Frequency Distributions
Observations that are measured on a numerical scale are referred to as quantitative
data. Weights, ages, bond yields, the length of a part in a manufacturing process, test
scores, and so on, are all examples of quantitative data. Out of the total population of
entering freshmen at a large university, a random sample of 100 students is selected,
and their entrance examination scores are recorded in Table 3.
Table 3 Entrance Examination Scores of 100 Entering Freshmen

762 451 602 440 570 553 367 520 454 653
433 508 520 603 532 673 480 592 565 662
712 415 595 580 643 542 470 743 608 503
566 493 635 780 537 622 463 613 502 577
618 581 644 605 588 695 517 537 552 682
340 537 370 745 605 673 487 412 613 470
548 627 576 637 787 507 566 628 676 750
442 591 735 523 518 612 589 648 662 512
663 588 627 584 672 533 738 455 512 622
544 462 730 576 588 705 695 541 537 563
The mass of raw data in Table 3 certainly does not elicit much interest or ex-
hibit much useful information. The data must be organized in some way so that it is
comprehensible. This can be done by constructing a frequency table. We generally
choose 5 to 20 class intervals of equal length to cover the data range—the more data,
the greater the number of intervals. Then we tally the data relative to these intervals.
The data range in Table 3 is 787 - 340 = 447 (found by subtracting the smallest
value in the data from the largest). If we choose 10 intervals, each of length 50, we
will be able to cover all the scores. Table 4 shows the result of this tally.
At first, it might seem appropriate to start at 300 and form the class intervals:
300–350, 350–400, 400–450, and so on. But if we do this, where will we place 350
or 400? We could, of course, adopt a convention of placing scores falling on an upper
boundary of a class in the next higher class (and some people do this); however, to
avoid confusion, we will always use one decimal place more for class boundaries than
what appears in the raw data. In this case, we chose the class intervals 299.5–349.5,
349.5–399.5, and so on, so that each score could be assigned to one and only one class
interval.
The number of measurements that fall within a given class interval is called
the class frequency, and the set of all such frequencies associated with their corre-
sponding classes is called a frequency distribution. Table 4 represents a frequency
Table 4 Frequency Table

Class Interval Tally Frequency Relative Frequency
299.5–349.5 ∙ 1 .01
349.5–399.5 ∙ ∙ 2 .02
399.5–449.5 ∙∙∙∙ 5 .05
449.5–499.5 ∙∙∙∙ ∙∙∙∙ 10 .10
499.5–549.5 ∙∙∙∙ ∙∙∙∙ ∙∙∙∙ ∙∙∙∙ ∙ 21 .21
549.5–599.5 ∙∙∙∙ ∙∙∙∙ ∙∙∙∙ ∙∙∙∙ 20 .20
599.5–649.5 ∙∙∙∙ ∙∙∙∙ ∙∙∙∙ ∙∙∙∙ 19 .19
649.5–699.5 ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ ∙ 11 .11
699.5–749.5 ∙ ∙ ∙ ∙ ∙ ∙ 7 .07
749.5–799.5 ∙ ∙ ∙ ∙ 4 .04
100 1.00
distribution of the set of raw scores in Table 3. If we divide each frequency by the
total number of items in the original data set (in our case 100), we obtain the relative
frequency of the data falling in each class interval—that is, the percentage of the
whole that falls in each class interval (see the last column in Table 4).
The relative frequencies also can be interpreted as probabilities associated with
the experiment, “A score is drawn at random out of the 100 in the sample.” An appro-
priate sample space for this experiment would be the set of simple outcomes
e1 = a score falls in the first class interval
e2 = a score falls in the second class interval
f
e10 = a score falls in the tenth class interval
The set of relative frequencies is then referred to as the probability distribution for
the sample space.
EXAMPLE 1 Determining Probabilities from a Frequency Table Referring to Table 4 and the
probability distribution just described, determine the probability that
(A) A randomly drawn score is between 499.5 and 549.5.
(B) A randomly drawn score is between 449.5 and 649.5.
SOLUTION
(A) Since the relative frequency associated with the class interval 499.5–549.5 is
.21, the probability that a randomly drawn score (from the sample of 100) will
fall in this interval is .21.
(B) Since a score falling in the interval 449.5–649.5 is a compound event, we sim-
ply add the probabilities for the simple events whose union is this compound
event. We add the probabilities corresponding to each class interval from 449.5
to 649.5 to obtain
.10 + .21 + .20 + .19 = .70
Matched Problem 1 Repeat Example 1 for the following intervals:
(A) 649.5–699.5 (B) 299.5–499.5
Comments on Statistics
POPULATION
(Set of all measurements Now, of course, what we are really interested in is whether the probability distribution
of interest to the sampler) for the sample of 100 entrance examination scores has anything to do with the total
population of entering freshmen at the university. This is a problem for the branch of
mathematics called statistics, which deals with the process of making inferences about
SAMPLE a total population based on random samples drawn from the population. Figure 10
(Subset of the
population) schematically illustrates the inferential statistical process. We will not go too far
into inferential statistics in this book since the subject is studied in detail in statistics
Figure 10 Inferential statistics: courses but our work in probability provides a good foundation for this study.
Based on information obtained from Intuitively, in the entrance examination example, we would expect that the larger
a sample, the goal of statistics is to the sample size, the more closely the probability distribution for the sample will ap-
make inferences about the population proximate that for the total population.
as a whole.
Histograms
A histogram is a special kind of vertical bar graph. In fact, if you rotate Table 4
counterclockwise 90°, the tally marks in the table take on the appearance of a bar
graph. Histograms have no space between the bars; class boundaries are located
on the horizontal axis, and frequencies are associated with the vertical axis.
Figure 11 is a histogram for the frequency distribution in Table 4. Note that we

have included both frequencies and relative frequencies on the vertical scale.
You can include either one or the other, or both, depending on what needs to be
emphasized. The histogram is the most common graphical representation of fre-
quency distributions.
Relative
Frequency frequency
20 .20
15 .15
10 .10
5 .05
0
0 299.5 399.5 499.5 599.5 699.5 799.5
349.5 449.5 549.5 649.5 749.5
Entrance examination scores
Figure 11 Histogram
EXAMPLE 2 Constructing Histograms with a Graphing Calculator Twenty vehicles were

chosen at random upon arrival at a vehicle emissions inspection station, and the
time elapsed (in minutes) from arrival to completion of the emissions test was re-
corded for each of the vehicles:
5 12 11 4 7 20 14 8 12 11
18 15 14 9 10 13 12 20 26 17
(A) Use a graphing calculator to draw a histogram of the data, choosing the five
class intervals 2.5–7.5, 7.5–12.5, and so on.
(B) What is the probability that for a vehicle chosen at random from the sample,
the time required at the inspection station will be less than 12.5 minutes? That
it will exceed 22.5 minutes?
SOLUTION
(A) Various kinds of statistical plots can be drawn by most graphing calculators. To
draw a histogram we enter the data as a list, specify a histogram from among the
various statistical plotting options, set the window variables, and graph. Figure 12
shows the data entered as a list, the settings of the window variables, and the result-
ing histogram for a particular graphing calculator. For details, consult your manual.
Figure 12
(B) From the histogram in Figure 12, we see that the first class has frequency 3 and
the second has frequency 8. The upper boundary of the second class is 12.5,
and the total number of data items is 20. Therefore, the probability that the time
required will be less than 12.5 minutes is
3 + 8 11
= = .55
20 20
Similarly, since the frequency of the last class is 1, the probability that the time
required will exceed 22.5 minutes is
1
= .05
20
Matched Problem 2 The weights (in pounds) were recorded for 20 kinder-
garten children chosen at random:
51 46 37 39 48 42 41 44 57 36
47 44 41 50 45 46 34 39 42 44
(A) Use a graphing calculator to draw a histogram of the data, choosing the five
class intervals 32.5–37.5, 37.5–42.5, and so on.
(B) What is the probability that a kindergarten child chosen at random from the
sample weighs less than 42.5 pounds? More than 42.5 pounds?
Frequency Polygons and Cumulative Frequency Polygons

A frequency polygon is a broken-line graph where successive midpoints of the tops
of the bars in a histogram are joined by straight lines. To draw a frequency polygon
for a frequency distribution, you do not need to draw a histogram first; you can just
locate the midpoints and join them with straight lines. Figure 13 is a frequency poly-
gon for the frequency distribution in Table 4. If the amount of data becomes very
large and we substantially increase the number of classes, then the frequency polygon
will take on the appearance of a smooth curve called a frequency curve.
Relative
Frequency frequency
25 .25
20 .20
15 .15
10 .10
5 .05
0
0 324.5 424.5 524.5 624.5 724.5 824.5
274.5 374.5 474.5 574.5 674.5 774.5
Figure 13 Frequency polygon
If we are interested in how many or what percentage of a total sample lies above
or below a particular measurement, a cumulative frequency table and polygon are
useful. Using the frequency distribution in Table 4, we accumulate the frequencies by
starting with the first class and adding frequencies as we move down the column. The
results are shown in Table 5.
Table 5 Cumulative Frequency Table

Class Interval Frequency Cumulative Frequency Relative Cumulative Frequency
299.5–349.5 1 1 .01
349.5–399.5 2 3 .03
399.5–449.5 5 8 .08
449.5–499.5 10 18 .18
499.5–549.5 21 39 .39
549.5–599.5 20 59 .59
599.5–649.5 19 78 .78
649.5–699.5 11 89 .89
699.5–749.5 7 96 .96
749.5–799.5 4 100 1.00
To form a cumulative frequency polygon, or ogive as it is also called, the cumula-

tive frequency is plotted over the upper boundary of the corresponding class. Figure 14
is the cumulative frequency polygon for the cumulative frequency table in Table 5.
Notice that we can easily see that 78% of the students scored below 649.5, while
only 18% scored below 499.5. We can conclude that the probability of a randomly
selected score from the sample of 100 lying below 649.5 is .78 and above 649.5 is
1.00 - .78 = .22.
Cumulative Relative cumulative
frequency frequency
100%
100 1.00
80 .80 78%
60 .60
40 .40
20 .20 18%
0
0 299.5 399.5 499.5 599.5 699.5 799.5
349.5 449.5 549.5 649.5 749.5
Figure 14 Cumulative frequency polygon (ogive)
CONCEPTUAL INSIGHT
Above each class interval in Figure 14, the cumulative frequency polygon is linear.
Such a function is said to be piecewise linear. The slope of each piece of a cumula-
tive frequency polygon is greater than or equal to zero (that is, the graph is never fall-
ing). The piece with the greatest slope corresponds to the class interval that has the
greatest frequency. In fact, for any class interval, the frequency is equal to the slope
of the cumulative frequency polygon multiplied by the width of the class interval.
Exercises 9.1
1. (A) Construct a frequency table and histogram for the (C) How are the histograms of parts (A) and (B) different?
following data set using a class interval width of 2, 3. The graphing calculator command shown in Figure A gener-
starting at 0.5. ated a set of 400 random integers from 2 to 24, stored as list
6 7 2 7 9 L 1. The statistical plot in Figure B is a histogram of L 1, using
6 4 7 6 6 a class interval width of 1, starting at 1.5.
(B) Construct a frequency table and histogram for the fol-
lowing data set using a class interval width of 2, starting
at 0.5.
5 6 8 1 3
5 10 7 6 8
(C) How are the two histograms of parts (A) and (B) similar?
How are the two data sets different?
2. (A) Construct a frequency table and histogram for the data (A) (B)
set of part (A) of Problem 1 using a class interval width
(A) Explain how the window variables can be changed to dis-
of 1, starting at 0.5.
play a histogram of the same data set using a class interval
(B) Construct a frequency table and histogram for the data width of 2, starting at 1.5. A width of 4, starting at 1.5.
set of part (B) of Problem 1 using a class interval width
(B) Describe the effect of increasing the class interval width
of 1, starting at 0.5.
on the shape of the histogram.
4. An experiment consists of rolling a pair of dodecahedral World Gold Production

(12-sided) dice and recording their sum (the sides of each die
are numbered from 1 to 12). The command shown in Figure Canada 2015
A simulated 500 rolls of the dodecahedral dice. The statistical 2010
plot in Figure B is a histogram of the 500 sums using a class
interval width of 1, starting at 1.5. United States
(A) Explain how the window variables can be changed to
display a histogram of the same data set using a class Russia
interval width of 2, starting at 1.5. A width of 3, starting
at - 0.5.
Australia
(B) Describe the effect of increasing the class interval width
on the shape of the histogram.
China
0 50 100 150 200 250 300 350 400 450

Gold production (metric tons)
8. Gasoline prices. Graph the data in the following table using

a divided bar graph. (Source: American Petroleum Institute)
State Gasoline Prices, April 2017

(A) (B) Price Before Tax
State ($ per gallon) Tax
California 2.41 .57
Applications
Hawaii 2.42 .63
Michigan 1.90 .59
New York 1.83 .62
5. Gross domestic product. Graph the data in the following Texas 1.80 .38
table using a bar graph. (Source: U.S. Bureau of Economic
Analysis)
9. Postal service. Graph the data in the following table using a
Gross Domestic Product (GDP) broken line graph. (Source: U.S. Postal Service)
Year GDP (trillion $)
U.S. Postal Service Employees
1975 5.49
1985 7.71 Year Number of Employees
1995 10.28 1990 760,668
2005 14.37 1995 753,384
2015 16.49 2000 787,538
2005 704,716
2010 583,908
6. Corporation revenues. Graph the data in the following
2015 491,863
table using a bar graph. (Source: Fortune)
Corporation Revenues, 2015 10. Postal service. Refer to Problem 9. If the data were
Corporation Revenue (billion $) presented in a bar graph, would horizontal bars or vertical
Walmart 482 bars be used? Could the data be presented in a pie graph?
ExxonMobil 246 Explain.
Apple 234 11. Federal income. Graph the data in the following table using
Berkshire Hathaway 211 a pie graph. (Source: U.S. Treasury)
McKesson 192
United Health 157 Federal Income by Source, 2015
Source Income (billion $)
7. Gold production. Use the double bar graph on world Personal income tax 1,365
gold production to determine the country that showed the Social insurance taxes 943
greatest increase in gold production from 2010 to 2015. Corporate income tax 292
Which country showed the greatest percentage increase?
Borrowing to cover deficit 390
The greatest percentage decrease? (Source: U.S. Geologi-
Other 260
cal Survey)
12. Gasoline prices. In April 2017, the average price of a gallon (A) Construct a frequency and relative frequency table using
of gasoline in the United States was $2.47. Of this amount, a class interval of 5, starting at - 0.5.
126 cents was the cost of crude oil, 57 cents the cost of refin- (B) Construct a histogram.
ing, 20 cents the cost of distribution and marketing, and 44
cents the amount of tax. Use a pie graph to present this data. (C) Construct a frequency polygon.
(Source: Commodity HQ.com)
(D) Construct a cumulative frequency and relative cumula-
13. Starting salaries. The starting salaries (in thousands of dol- tive frequency table. What is the probability that a price–
lars) of 20 graduates, chosen at random from the graduating earnings ratio drawn at random from the sample will fall
class of an urban university, were determined and recorded in between 4.5 and 14.5?
the following table:
(E) Construct a cumulative frequency polygon.
Starting Salaries
44 39 37 49 51 16. Mouse weights. One hundred healthy mice were weighed at
38 42 47 45 46
the beginning of an experiment with the following results:
33 41 43 44 39 Mouse Weights (grams)
37 45 39 40 42
51 54 47 53 59 46 50 50 56 46
(A) Construct a frequency and relative frequency table using 48 50 45 49 52 55 42 57 45 51
a class interval width of 4 starting at 30.5. 53 55 51 47 53 53 49 51 43 48
44 48 54 46 49 51 52 50 55 51
(B) Construct a histogram. 50 53 45 49 57 54 53 49 46 48
(C) What is the probability that a graduate chosen from the 52 48 50 52 47 50 44 46 47 49
sample will have a starting salary above $42,500? Below 49 51 57 49 51 42 49 53 44 52
$38,500? 53 55 48 52 44 46 54 54 57 55
48 50 50 55 52 48 47 52 55 50
(D) Construct a histogram using a graphing calculator.
59 52 47 46 56 54 51 56 54 55
14. Commute times. Thirty-two people were chosen at random
(A) Construct a frequency and relative frequency table using
from among the employees of a large corporation. Their com-
a class interval of 2, starting at 41.5.
mute times (in hours) from home to work were recorded in
the following table: (B) Construct a histogram.
Commute Times (C) Construct a frequency polygon.
0.5 0.9 0.2 0.4 0.7 1.2 1.1 0.7
0.6 0.4 0.8 1.1 0.9 0.3 0.4 1.0 (D) Construct a cumulative frequency and relative cumula-
0.9 1.0 0.7 0.3 0.6 1.1 0.7 1.1
tive frequency table. What is the probability of a mouse
weight drawn at random from the sample lying between
0.4 1.3 0.7 0.6 1.0 0.8 0.4 0.9
45.5 and 53.5?
a class interval width of 0.2, starting at 0.15. (E) Construct a cumulative frequency polygon.
(B) Construct a histogram. 17. Population growth. Graph the data in the following table
using a broken-line graph.
(C) What is the probability that a person chosen at random
from the sample will have a commuting time of at least Annual World Population Growth
an hour? Of at most half an hour? Year Growth (millions)
(D) Construct a histogram using a graphing calculator. 1900 11

1925 21
15. Common stocks. The following table shows price–earnings 1950 46
ratios of 100 common stocks chosen at random from the New 1975 71
York Stock Exchange. 2000 80
Price–Earnings (PE) Ratios 2025 71
7 11 6 6 10 6 31 28 13 19
18. AIDS epidemic. One way to gauge the toll of the AIDS
6 18 9 7 5 5 9 8 10 6
epidemic in Sub-Saharan Africa is to compare life expec-
10 3 4 6 7 9 9 19 7 9
tancies with the figures that would have been projected in
17 33 17 12 7 5 7 10 7 9
the absence of AIDS. Use the broken-line graphs shown to
18 17 4 6 11 13 7 6 10 7
estimate the life expectancy of a child born in the year 2012.
7 9 8 15 16 11 10 7 5 14 What would the life expectancy of the same child be in the
12 10 6 7 7 13 10 5 6 4 absence of AIDS? For which years of birth is the life expec-
10 6 7 11 19 17 6 9 6 5 tancy less than 50 years? If there were no AIDS epidemic, for
6 13 4 7 6 12 9 14 9 7 which years of birth would the life expectancy be less than
18 5 12 8 8 8 13 9 13 15 50 years?
Life Expectancy in Sub-Saharan Africa 24. Study abroad. Would a pie graph be more effective or less
70 effective than the bar graph shown in presenting information
on the most popular destinations of U.S. college students
60 Without AIDS who study abroad? Justify your answer. (Source: Institute
With AIDS of International Education)
Life expectancy (years)
50
Destinations of U.S. Students Studying Abroad, 2014
40
United Kingdom
30 Italy
Spain
20
France
10 China
Germany
1985–90 1990–95 1995–00 2000–05 2005–10 2010–15
Ireland
Year of birth
Figure for 18 Costa Rica
19. Nutrition. Graph the data in the following table using a Australia
double bar graph. Japan
0 5 10 15 20 25 30 35 40
Recommended Daily Allowances Number of students (thousands)
Males Females
25. Median age. Use the broken-line graph shown to estimate the
Grams of Age 15–18 Age 15–18 median age in 1900 and 2000. In which decades did the me-
Carbohydrate 375 275 dian age increase? In which did it decrease? Discuss the factors
Protein 60 44 that may have contributed to the increases and decreases.
Fat 100 73 Median Age in the United States, 1900–2010
40
20. Greenhouse gases. The U.S. Environmental Protection 35
Agency estimated that of all emissions of greenhouse gases
by the United States in 2014, carbon dioxide accounted for 30
81%, methane for 11%, nitrous oxide for 6%, and fluorinated 25
gases for 3%. Use a pie graph to present this data. Find the 20
central angles of the graph.
15
21. Nutrition. Graph the nutritional information in the following
10
table using a double bar graph.
5
Fast-Food Burgers: Nutritional Information
0
Calories Calories from Fat 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
2-oz burger, plain 268 108 26. State prisoners. In 1980 in the United States, 6% of the
2 addl. oz of beef 154 90 inmates of state prisons were incarcerated for drug offenses,
1 slice cheese 105 81 30% for property crimes, 4% for public order offenses, and
3 slices bacon 109 81 60% for violent crimes; in 2014 the percentages were 16%,
1 tbsp. mayonnaise 100 99 19%, 12%, and 53%, respectively. Present the data using two
pie graphs. Discuss factors that may account for the shift in
22. Nutrition. Refer to Problem 21. Suppose that you are trying percentages between 1980 and 2014. (Source: U.S. Depart-
to limit the fat in your diet to at most 30% of your calories, ment of Justice)
and your calories down to 2000 per day. Should you order 27. Grade-point averages. One hundred seniors were chosen
the quarter-pound bacon cheeseburger with mayo for lunch? at random from a graduating class at a university and their
How would such a lunch affect your choice of breakfast and grade-point averages recorded:
dinner? Discuss.
23. Education. For statistical studies, U.S. states are often
a class interval of 0.2 starting at 1.95.
grouped by region: Northeast, Midwest, South, and West. The
1965 total public school enrollment (in millions) in each re- (B) Construct a histogram.
gion was 8.8, 11.8, 13.8, and 7.6, respectively. The projected
(C) Construct a frequency polygon.
2020 enrollment was 7.8, 10.8, 20.4, and 13.6, respectively.
Use two pie graphs to present this data and discuss any trends (D) Construct a cumulative frequency and relative cumula-
suggested by your graphs, (Source: U.S. Department of tive frequency table. What is the probability of a GPA
Education) drawn at random from the sample being over 2.95?
(E) Construct a cumulative frequency polygon. Answers to Matched Problems

Grade-Point Averages (GPA) 1. (A) .11 (B) .18
2.1 2.0 2.7 2.6 2.1 3.5 3.1 2.1 2.2 2.9
2. (A) (B) .45; .55
2.3 2.5 3.1 2.2 2.2 2.0 2.3 2.5 2.1 2.4
2.7 2.9 2.1 2.2 2.5 2.3 2.1 2.1 3.3 2.1
2.2 2.2 2.5 2.3 2.7 2.4 2.8 3.1 2.0 2.3
2.6 3.2 2.2 2.5 3.6 2.3 2.4 3.7 2.5 2.4
3.5 2.4 2.3 3.9 2.9 2.7 2.6 2.1 2.4 2.0
2.4 3.3 3.1 2.8 2.3 2.5 2.1 3.0 2.6 2.3
2.1 2.6 2.2 3.2 2.7 2.8 3.4 2.7 3.6 2.1
2.7 2.8 3.5 2.4 2.3 2.0 2.1 3.1 2.8 2.1
3.8 2.5 2.7 2.1 2.2 2.4 2.9 3.3 2.0 2.6
9.2 Measures of Central Tendency

■■ Mean In Section 9.1 we saw that graphical techniques contributed substantially to under-
■■ Median
standing large masses of raw data. In this and the next section, we discuss several
important numerical measures that are used to describe sets of data. These numerical
■■ Mode descriptions are generally of two types:
1. Measures that indicate the approximate center of a distribution, called measures
of central tendency; and
2. Measures that indicate the amount of scatter about a central point, called measures
of dispersion.
In this section, we look at three widely used measures of central tendency, and in
the next section we consider measures of dispersion.
Mean
If a student’s scores on four exams are x1 = 82, x2 = 75, x3 = 86, and x4 = 87,
then the mean, or arithmetic average, of the scores is the sum of the scores divided
by the number of scores:
x1 + x2 + x3 + x4 82 + 75 + 86 + 87 330
= = = 82.5
4 4 4
the sum of the scores. We use the summation symbol g (the uppercase Greek letter
It is convenient, especially when there are many scores, to use a compact notation for
sigma) to indicate that the scores should be added (see Appendix C.1). So
a xi denotes x1 + x2 + x3 + x4
4
i=1
and the mean of the four exam scores is
a xi
4
i=1
= 82.5
4
The mean is a single number that, in a sense, represents the entire data set. It in-
volves all the measurements in the set and it is easily computed. Because of these and
other desirable properties, the mean is the most widely used measure of central tendency.
In statistics, we are concerned with both a sample mean and the mean of the cor-
responding population (the sample mean is often used as an estimator for the popula-
tion mean), so it is important to use different symbols to represent these two means. It
is customary to use a letter with an overbar, such as x, to represent a sample mean and
the Greek letter m (“mu”) to represent a population mean.
x = sample mean m = population mean
SECTION 9.2 Measures of Central Tendency 491
DEFINITION Mean: Ungrouped Data

If x1, x2,c, xn is a set of n measurements, then the mean of the set of measure-
ments is given by
a xi
n
i=1 x1 + x2 + g + xn
3mean4 = = (1)
n n
where
x = 3mean4 if data set is a sample
m = 3mean4 if data set is the population
EXAMPLE 1 Finding the Mean Find the mean for the sample measurements 3, 5, 1, 8, 6, 5, 4,
and 6.
SOLUTION Solve using formula (1):
a xi
n
i=1 3 + 5 + 1 + 8 + 6 + 5 + 4 + 6 38
x = = = = 4.75
n 8 8
Matched Problem 1 Find the mean for the sample measurements 3.2, 4.5,
2.8, 5.0, and 3.6.
Fifteen customers filled out a satisfaction survey, rating the service they received
by checking a number from 1 to 5. Here are the ratings:
5, 3, 5, 4, 4, 3, 4, 5, 5, 5, 4, 2, 5, 5, 4
To find the sum, we could simply add the 15 numbers. But there is another option.
Note that the score 5 appears seven times, 4 appears five times, 3 appears twice, 2
appears once, and 1 appears zero times. The sum of all 15 scores can therefore be
obtained by multiplying each possible score xi by its frequency fi:
a xi fi = 5172 + 4152 + 3122 + 2112 + 1102 = 63

5
i=1
The mean of the 15 scores is
a xi fi
5
i=1 63
= = 4.2
15 15
For grouped data, such as in Table 4 on page 482, we may not have the option of
simply adding the scores; the original data set may be unavailable. Instead, we form a
sum similar to that of the customer satisfaction survey, where each term is an xi mul-
tiplied by its frequency fi, as explained in the definition of the mean for grouped data.
DEFINITION Mean: Grouped Data

A data set of n measurements is grouped into k classes in a frequency table. If xi is
the midpoint of the ith class interval and fi is the ith class frequency, then the mean
for the grouped data is given by
a xi fi
k
i=1 x1 f1 + x2 f2 + g + xk fk
3mean4 = = (2)
n n
where
n = a fi = total number of measurements

k
i=1
x = 3mean4 if data set is a sample

m = 3mean4 if data set is the population
Note that n is the total number of measurements in the entire data

! CAUTION set—not the number of classes!
The mean computed by formula (2) is a weighted average of the midpoints of

the class intervals. In general, this will be close to, but not exactly the same as, the
mean computed by formula (1) for ungrouped data.
EXAMPLE 2 Finding the Mean for Grouped Data Find the mean for the sample data sum-
marized in Table 4 on page 482.
SOLUTION In Table 1, we repeat part of Table 4, adding columns for the class mid-
points xi and the products xi fi.
Table 1 Entrance Examination Scores

Class Interval Midpoint xi Frequency fi Product xi fi
299.5–349.5 324.5 1 324.5
349.5–399.5 374.5 2 749.0
399.5–449.5 424.5 5 2,122.5
449.5–499.5 474.5 10 4,745.0
499.5–549.5 524.5 21 11,014.5
549.5–599.5 574.5 20 11,490.0
599.5–649.5 624.5 19 11,865.5
649.5–699.5 674.5 11 7,419.5
699.5–749.5 724.5 7 5,071.5
749.5–799.5 774.5 4 3,098.0
n = a fi = 100 a xi fi = 57,900.0
10 10
i=1 i=1
The average entrance examination score for the sample of 100 entering freshmen is
a xi fi
k
i=1 57,900
579 x = = = 579
n 100
If the histogram for the data in Table 1 (Fig. 11, Section 9.1) was drawn on
Figure 1 The balance point on the
a piece of wood of uniform thickness and the wood cut around the outside of the
histogram is x = 579. figure, then the resulting object would balance exactly at the mean x = 579, as
shown in Figure 1.
Matched Problem 2 Compute the mean for the grouped sample data listed in
Table 2.
Table 2
Class Interval Frequency
0.5–5.5 6
5.5–10.5 20
10.5–15.5 18
15.5–20.5 4
CONCEPTUAL INSIGHT
The mean for ungrouped data and the mean for grouped data can be interpreted as
the expected values of appropriately chosen random variables (see Section 7.5).
Consider a set of n measurements x1, x2, c, xn (ungrouped data). Let S
be the sample space consisting of n simple events (the n measurements), each
equally likely. Let X be the random variable that assigns the numerical value xi
to each simple event in S. Then each measurement xi has probability pi = 1n. The
expected value of X is given by
E1X2 = x1p1 + x2p2 + g + xnpn
= x1 # + x2 # + g + xn #
1 1 1
n n n
x1 + x2 + g + xn
=
n
= 3mean4
Similarly, consider a set of n measurements grouped into k classes in a fre-
quency table (grouped data). Let S′ be the sample space consisting of n simple
events (the n measurements), each equally likely. Let X′ be the random variable
that assigns the midpoint xi of the ith class interval to the measurements that belong
to that class interval. Then each midpoint xi has probability pi = fi # 1n, where fi
denotes the frequency of the ith class interval. The expected value of X′ is given by
E1X′2 = x1 p1 + x2 p2 + g + xk pk
= x1 af1 # b + x2 af2 # b + g + xk afk # b

1 1 1
n n n
x1 f1 + x2 f2 + g + xk fk
=
n
= 3mean4
Median
Occasionally, the mean can be misleading as a measure of central tendency. Suppose
the annual salaries of seven people in a small company are $34,000, $40,000,
$56,000, $36,000, $36,000, $156,000, and $48,000. The mean salary is
a xi
n
i=1 $406,000
x = = = $58,000
n 7
Six of the seven salaries are below the average! The one large salary distorts the results.
A measure of central tendency that is not influenced by extreme values is the
median. The following definition of median makes precise our intuitive notion
of the “middle element” when a set of measurements is arranged in ascending or
descending order. Some sets of measurements, for example, 5, 7, 8, 13, 21, have a
middle element. Other sets, for example, 9, 10, 15, 20, 23, 24, have no middle ele-
ment, or you might prefer to say that they have two middle elements. For any number
between 15 and 20, half the measurements fall above the number and half fall below.
DEFINITION Median: Ungrouped Data

1. If the number of measurements in a set is odd, the median is the middle measure-
ment when the measurements are arranged in ascending or descending order.
2. If the number of measurements in a set is even, the median is the mean of the
two middle measurements when the measurements are arranged in ascending
or descending order.
EXAMPLE 3 Finding the Median Find the median salary in the preceding list of seven salaries.
SOLUTION Arrange the salaries in increasing order and choose the middle one:
Salary
$34,000
36,000
36,000
40,000 — Median ($40,000)
48,000
56,000 — Mean ($58,000)
156,000
In this case, the median is a better measure of central tendency than the mean.
Matched Problem 3 Add the salary $100,000 to those in Example 3 and com-
pute the median and mean for these eight salaries.
The median, as we have defined it, is easy to determine and is not influenced by
extreme values. Our definition does have some minor handicaps, however. First, if
the measurements we are analyzing were carried out in a laboratory and presented
to us in a frequency table, we may not have access to the individual measurements.
In that case we would not be able to compute the median using the above definition.
Second, a set like 4, 4, 6, 7, 7, 7, 9 would have median 7 by our definition, but 7 does
not possess the symmetry we expect of a “middle element” since there are three mea-
surements below 7 but only one above.
To overcome these handicaps, we define a second concept, the median for
grouped data. To guarantee that the median for grouped data exists and is unique, we
assume that the frequency table for the grouped data has no classes of frequency 0.
DEFINITION Median: Grouped Data

The median for grouped data with no classes of frequency 0 is the number such
Figure 2 The area to the left of that the histogram has the same area to the left of the median as to the right of the
the median equals the area to the median (see Fig. 2).
right.
EXAMPLE 4 Finding the Median for Grouped Data Compute the median for the grouped
data in Table 3.
SOLUTION First we draw the histogram of the data (Fig. 3). The total area of the
histogram is 15, which is just the sum of the frequencies, since all rectangles have a
base of length 1. The area to the left of the median must be half the total area—that
is, 15
2 = 7.5. Looking at Figure 3, we see that the median M lies between 6.5 and
7.5. Thus, the area to the left of M, which is the sum of the blue shaded areas in
Figure 3, must be 7.5:
112132 + 112112 + 112122 + 1M - 6.52142 = 7.5
Solving for M gives M = 6.875. The median for the grouped data in Table 3 is 6.875.
3
Table 3
Class Interval Frequency 2
3.5–4.5 3
4.5–5.5 1
1
5.5–6.5 2
6.5–7.5 4
7.5–8.5 3 4 5 6 7 8 9
8.5–9.5 2 Figure 3
The area of the leftmost rectangle in Figure 3 is 112132 = 3

! CAUTION because the base of the rectangle is 1 and the height is 3. The
base is equal to the width of the class interval and the height is the frequency of that
class. See Matched Problem 4.
Matched Problem 4 Find the median for the grouped data in the following
table:
Class Interval Frequency

3.5–5.5 4
5.5–7.5 2
7.5–9.5 3
9.5–11.5 5
11.5–13.5 4
13.5–15.5 3
Mode
A third measure of central tendency is the mode.
DEFINITION Mode
The mode is the most frequently occurring measurement in a data set. There may
be a unique mode, several modes, or, if no measurement occurs more than once,
essentially no mode.
EXAMPLE 5 Finding Mode, Median, and Mean

Data Set Mode Median Mean
(A) 4, 5, 5, 5, 6, 6, 7, 8, 12 5 6 6.44
(B) 1, 2, 3, 3, 3, 5, 6, 7, 7, 7, 23 3, 7 5 6.09
(C) 1, 3, 5, 6, 7, 9, 11, 15, 16 None 7 8.11
Data set (B) in Example 5 is referred to as bimodal, since there are two modes. Since
no measurement in data set (C) occurs more than once, we say that it has no mode.
Matched Problem 5 Compute the mode(s), median, and mean for each data set:
(A) 2, 1, 2, 1, 1, 5, 1, 9, 4 (B) 2, 5, 1, 4, 9, 8, 7
(C) 8, 2, 6, 8, 3, 3, 1, 5, 1, 8, 3
The mode, median, and mean can be computed in various ways with the aid of
a graphing calculator. In Figure 4A, the data set of Example 5B is entered as a list,
and its median and mean are computed. The histogram in Figure 4B shows the two
modes of the same data set.
Figure 4 (A) (B)
As with the median, the mode is not influenced by extreme values. Suppose that in
the data set of Example 5B, we replace 23 with 8. The modes remain 3 and 7 and the me-
dian is still 5; however, the mean changes to 4.73. The mode is most useful for large data
sets because it emphasizes data concentration. For example, a clothing retailer would be
interested in the mode of sizes due to customer demand of various items in a store.
The mode also can be used for qualitative attributes—that is, attributes that are not
numerical. The mean and median are not suitable in these cases. For example, the mode
can be used to give an indication of a favorite brand of ice cream or the worst movie of
the year. Figure 5 shows the results of a random survey of 1,000 people on entree pref-
erences when eating dinner out. According to this survey, we would say that the modal
preference is beef. Note that the mode is the only measure of central tendency (loca-
tion) that can be used for this type of data; mean and median make no sense.
400
368 Mode
350
300 289
250
200
165
150
104
100
74
50
0
Fish Shellfish Beef Lamb Pork
Figure 5 The modal preference for an entree is beef.
In actual practice, the mean is used the most, the median next, and the mode a
distant third.

For many sets of measurements the median lies between the mode and the mean. But
this is not always so.
(A) In a class of seven students, the scores on an exam were 52, 89, 89, 92, 93, 96,
99. Show that the mean is less than the mode and that the mode is less than the
median.
(B) Construct hypothetical sets of exam scores to show that all possible orders
among the mean, median, and mode can occur.
Exercises 9.2
Skills Warm-up Exercises 12. Number
Car Color Preferring
W In Problems 1–4, find the mean of the data set. (If necessary,
review Section C.1). Red 1,324
White 3,084
1. 5, 8, 6, 7, 9
Black 1,617
2. 12, 15, 18, 21 Blue 2,303
Brown 2,718
3. - 1, - 2, 0, 2, 4
Gold 1,992
4. 85, 75, 65, 75
In Problems 5–8, find the indicated sum. (If necessary, review Find the mean for the sets of grouped data in Problems 13 and 14.
Section C.1).
13. Interval Frequency
5. a xi if xi = i + 3 for i = 1, 2, c, 6
6
0.5–2.5 2
i=1 2.5–4.5 5
6. a xi if xi = 100i for i = 1, 2, c, 5
5 4.5–6.5 7
i=1
6.5–8.5 1
7. a xifi if xi = 2i and fi = 3i for i = 6, 7, c, 10

10
14. Interval Frequency
i=6
0.5–2.5 5
8. a xifi if xi = i and fi = i for i = 5, 6, 7, 8
8
2 2.5–4.5 1
i=5 4.5–6.5 2
6.5–8.5 7
A Find the mean, median, and mode for the sets of ungrouped data
given in Problems 9 and 10.
B 15. Which single measure of central tendency—mean, me-
9. 1, 2, 2, 3, 3, 3, 3, 4, 4, 5 dian, or mode—would you say best describes the following
set of measurements? Discuss the factors that justify your
10. 1, 1, 1, 1, 2, 3, 4, 5, 5, 5
preference.
Find the mean, median, and/or mode, whichever are applicable, 8.01 7.91 8.13 6.24 7.95
in Problems 11 and 12. 8.04 7.99 8.09 6.24 81.2
11. Flavor Number Preferring
16. Which single measure of central tendency—mean, me-
Vanilla 139 dian, or mode—would you say best describes the following
Chocolate 376 set of measurements? Discuss the factors that justify your
Strawberry 89 preference.
Pistachio 105
Cherry 63 47 51 80 91 85
Almond mocha 228 69 91 95 81 60
17. A data set is formed by recording the results of 100 rolls of a 23. Lightbulb lifetime. Find the mean and median for the data in
fair die. the following table.
(A) What would you expect the mean of the data set to be?
The median? Life (Hours) of 50 Randomly Selected Lightbulbs
Interval Frequency
(B) Form such a data set by using a graphing calculator to
simulate 100 rolls of a fair die, and find its mean and 799.5–899.5 3
median. 899.5–999.5 10
999.5–1,099.5 24
18. A data set is formed by recording the sums on 200 rolls of a
pair of fair dice. 1,099.5–1,199.5 12
1,199.5–1.299.5 1
(A) What would you expect the mean of the data set to be?
The median? 24. Price–earnings ratios. Find the mean and median for the
(B) Form such a data set by using a graphing calculator data in the following table.
to simulate 200 rolls of a pair of fair dice, and find the
mean and median of the set. Price–Earnings Ratios of 100 Randomly Chosen
Stocks from the New York Stock Exchange
C 19. (A) Construct a set of four numbers that has mean 300,
Interval Frequency
median 250, and mode 175.
-0.5–4.5 5
(B) Let m1 7 m2 7 m3. Devise and discuss a procedure for
4.5–9.5 54
constructing a set of four numbers that has mean m1,
9.5–14.5 25
median m2, and mode m3.
14.5–19.5 9
20. (A) Construct a set of five numbers that has mean 200, me- 19.5–24.5 4
dian 150, and mode 50. 24.5–29.5 1
(B) Let m1 7 m2 7 m3. Devise and discuss a procedure 29.5–34.5 2
for constructing a set of five numbers that has mean m1,
median m2, and mode m3. 25. Student loan debt. Find the mean, median, and mode for the
data in the following table that gives the percentages by state of
students graduating from college in 2015 who had student loan
debt. (Source: The Institute for College Access & Success)
Applications College Graduates with Student Loan Debt

Percentage of 2015 College
21. Price–earnings ratios. Find the mean, median, and mode State Graduates with Student Loan Debt
for the data in the following table.
Arizona 56
Arkansas 57
Price–Earnings Ratios for Eight Stocks in a Portfolio
Louisiana 51
5.3 10.1 18.7 35.5
New Mexico 58
12.9 8.4 16.2 10.1
Oklahoma 52
Texas 56
22. Gasoline tax. Find the mean, median, and mode for the
data in the following table. (Source: American Petroleum 26. Tourism. Find the mean, median, and mode for the data in
Institute.) the following table. (Source: World Tourism Organization)
State Gasoline Tax, 2017 International Tourism Receipts, 2015

State Tax (cents per gal.) Country Receipts (billion $)
Wisconsin 32.9 United States 178.3
New York 43.9 China 114.1
Connecticut 39.9 Spain 56.5
Nebraska 28.2 France 45.9
Kansas 24.0 Thailand 44.6
Texas 20.0 United Kingdom 42.4
California 38.1 Italy 39.7
Tennessee 21.4 Germany 36.9
27. Mouse weights. Find the mean and median for the data in 30. Grade-point averages. Find the mean and median for the
the following table. grouped data in the following table.
Mouse Weights (grams) Graduating Class Grade-Point

Interval Frequency Averages
41.5–43.5 3 Interval Frequency
43.5–45.5 7 1.95–2.15 21
45.5–47.5 13 2.15–2.35 19
47.5–49.5 17 2.35–2.55 17
49.5–51.5 19 2.55–2.75 14
51.5–53.5 17 2.75–2.95 9
53.5–55.5 15 2.95–3.15 6
55.5–57.5 7 3.15–3.35 5
57.5–59.5 2 3.35–3.55 4
3.55–3.75 3
28. Blood cholesterol levels. Find the mean and median for the 3.75–3.95 2
data in the following table.
31. Entrance examination scores. Compute the median for the
Blood Cholesterol Levels grouped data of entrance examination scores given in Table 1
(milligrams per deciliter) on page 492.
Interval Frequency
32. Presidents. Find the mean and median for the grouped data
149.5–169.5 4 in the following table.
169.5–189.5 11
189.5–209.5 15 U.S. Presidents’ Ages at
209.5–229.5 25 Inauguration
229.5–249.5 13
Age Number
249.5–269.5 7
39.5–44.5 2
269.5–289.5 3
44.5–49.5 7
289.5–309.5 2
49.5–54.5 12
54.5–59.5 13
29. Immigration. Find the mean, median, and mode for the 59.5–64.5 7
data in the following table. (Source: U.S. Census 64.5–69.5 2
Bureau) 69.5–74.5 2
Top Ten Countries of Birth of U.S.

Foreign-Born Population, 2015
Number Answers to Matched Problems
Country (thousands) 1. x = 3.82
Mexico 11,643 2. x ≈ 10.1
China 2,677 3. Median = $44,000; mean = $63,250
India 2,390 4. Median for grouped data = 10.1
Philippines 1,982 5. Arrange each set of data in ascending order:
El Salvador 1,352
Vietnam 1,301 Data Set Mode Median Mean
Cuba 1,211 (A) 1, 1, 1, 1, 2, 2, 4, 5, 9 1 2 2.89
Dominican Republic 1,063 (B) 1, 2, 4, 5, 7, 8, 9 None 5 5.14
Korea 1,060 (C) 1, 1, 2, 3, 3, 3, 5, 6, 8, 8, 8 3, 8 3 4.36
Guatemala 928
9.3 Measures of Dispersion

■■ Range A measure of central tendency gives us a typical value that can be used to describe a
■■ Standard Deviation: Ungrouped
whole set of data, but this measure does not tell us whether the data are tightly clus-
Data tered or widely dispersed. We now consider two measures of variation—range and
standard deviation—that will give some indication of data scatter.
■■ Standard Deviation: Grouped Data
■■ Significance of Standard Deviation Range
A measure of dispersion, or scatter, that is easy to compute and can be easily under-
stood is the range. The range for a set of ungrouped data is the difference between
4 the largest and the smallest values in the data set. The range for a frequency dis-
tribution is the difference between the upper boundary of the highest class and the
2 2 lower boundary of the lowest class.
1 1 Consider the histograms in Figure 1. We see that the range adds only a little in-
formation about the amount of variation in a data set. The graphs clearly show that
8 9 10 11 12 even though each data set has the same mean and range, all three sets differ in the
(A) Mean 5 10 amount of scatter, or variation, of the data relative to the mean. The data set in part
Range 5 12.5 2 7.5 5 5 (A) is tightly clustered about the mean; the data set in part (B) is dispersed away from
the mean; and the data set in part (C) is uniformly distributed over its range.
4 4 Since the range depends only on the extreme values of the data, it does not give
us any information about the dispersion of the data between these extremes. We need
a measure of dispersion that will give us some idea of how the data are clustered or
1 1 scattered relative to the mean. Standard deviation is such a measure.
0
8 9 10 11 12
Standard Deviation: Ungrouped Data
(B) Mean 5 10 We will develop the concepts of variance and standard deviation—both measures of
Range 5 5 variation—through a simple example. Suppose that a random sample of five stamped
parts is selected from a manufacturing process, and these parts are found to have the
2 2 2 2 2
following lengths (in centimeters):
5.2, 5.3, 5.2, 5.5, 5.3
Computing the sample mean, we obtain
8 9 10 11 12
a xi
(C) Mean 5 10 n
Range 5 5
i=1 5.2 + 5.3 + 5.2 + 5.5 + 5.3
x = =
Figure 1 n 5
= 5.3 centimeters
Table 1 How much variation exists between the sample mean and all measurements in the
sample? As a first attempt at measuring the variation, let us represent the deviation of a
xi 1xi - x2
measurement from the mean by 1xi - x2. Table 1 lists all the deviations for this sample.
5.2 -0.1
Using these deviations, what kind of formula can we find that will give us a
5.3 0.0
single measure of variation? It appears that the average of the deviations might be a
5.2 -0.1
good measure. But look at what happens when we add the second column in Table
5.5 0.2
1. We get 0! It turns out that this will always happen for any data set. Now what?
5.3 0.0
We could take the average of the absolute values of the deviations; however, this ap-
proach leads to problems relative to statistical inference. Instead, to get around the
sign problem, we will take the average of the squares of the deviations and call this
number the variance of the data set:
a 1xi - x2
n
2
i=1
3variance4 = (1)
n
SECTION 9.3 Measures of Dispersion 501
Calculating the variance using the entries in Table 1, we have
a 1xi - 5.32
5
2
i=1
3variance4 = = 0.012 square centimeter
5
We still have a problem because the units in the variance are square centimeters
instead of centimeters (the units of the original data set). To obtain the units of the
original data set, we take the positive square root of the variance and call the result
the standard deviation of the data set:
a 1xi - x2 a 1xi - 5.32

n 5
2 2
i=1 i=1
3standard deviation4 = = (2)
H n H 5
≈ 0.11 centimeter
The sample variance is usually denoted by s2 and the population variance by s2
(s is the Greek lowercase letter “sigma”). The sample standard deviation is usually
denoted by s and the population standard deviation by s.
In inferential statistics, the sample variance s2 is often used as an estimator for
the population variance s2 and the sample standard deviation s for the population
standard deviation s. It can be shown that one can obtain better estimates of the pop-
ulation parameters in terms of the sample parameters (particularly when using small
samples) if the divisor n is replaced by n - 1 when computing sample variances or
sample standard deviations.
DEFINITION Variance: Ungrouped Data*

The sample variance s2 of a set of n sample measurements x1, x2,c, xn with mean
x is given by
a 1xi - x2
n
2
i=1
s2 = (3)
n - 1
If x1, x2,c, xn is the whole population with mean m, then the population variance
s2 is given by
a 1xi - m2
n
2
i=1
s2 =
n
*
In this section, we restrict our interest to the sample variance.
The standard deviation is just the positive square root of the variance.
DEFINITION Standard Deviation: Ungrouped Data†

The sample standard deviation s of a set of n sample measurements x1, x2,c, xn
with mean x is given by
a 1xi - x2
n
2
i=1
s = (4)
H n - 1
If x1, x2,c, xn is the whole population with mean m, then the population standard
deviation s is given by
a 1xi - m2
n
2
i=1
s =
H n
†
In this section, we restrict our interest to the sample standard deviation.
Computing the standard deviation for the original sample measurements (see
Table 1), we now obtain
a 1xi - 5.32
5
2
i=1
s = ≈ 0.12 cm
H 5 - 1
EXAMPLE 1 Finding the Standard Deviation Find the standard deviation for the sample mea-
surements 1, 3, 5, 4, 3.
SOLUTION To find the standard deviation for the data set, we can utilize a table or
use a calculator. Most will prefer the latter. Here is what we compute:
1 + 3 + 5 + 4 + 3
x = = 3.2
5
11 - 3.22 2 + 13 - 3.22 2 + 15 - 3.22 2 + 14 - 3.22 2 + 13 - 3.22 2
s =
A 5 - 1
≈ 1.48
Matched Problem 1 Find the standard deviation for the sample measure-
ments 1.2, 1.4, 1.7, 1.3, 1.5.
Remark—Many graphing calculators can compute x and s directly after the sample
measurements are entered—a helpful feature, especially when the sample is fairly
large. This shortcut is illustrated in Figure 2 for a particular graphing calculator, where
the data from Example 1 are entered as a list, and several different one-variable sta-
tistics are immediately calculated. Included among these statistics are the mean x, the
sample standard deviation s (denoted by Sx in Fig. 2B), the population standard devia-
tion s (denoted by sx), the number n of measurements, the smallest element of the
data set (denoted by minX), the largest element of the data set (denoted by maxX), the
median (denoted by Med), and several statistics we have not discussed.
(A) Data (B) Statistics (C) Statistics (continued)
Figure 2
CONCEPTUAL INSIGHT
If the sample measurements in Example 1 are considered to constitute the whole
population, then the population standard deviation sx is approximately equal to 1.33
[see Fig. 2(B)]. The computation of sx is the same as that of Example 1, except that
the denominator n - 1 1 = 5 - 12 under the radical sign is replaced by n 1 = 52.
Consequently, Sx ≈ 1.48 is greater than sx ≈ 1.33. Formulas (4) and (2) produce
nearly the same results when the sample size n is large. The law of large numbers
states that we can make a sample standard deviation s as close to the population
standard deviation s as we like by making the sample sufficiently large.

(A) When is the sample standard deviation of a set of measurements equal to 0?
(B) Can the population standard deviation of a set of measurements ever be greater
than the range? Explain why or why not.
Standard Deviation: Grouped Data

Formula (4) for sample standard deviation is extended to grouped sample data as de-
scribed in the following box:
DEFINITION Standard Deviation: Grouped Data*

Suppose a data set of n sample measurements is grouped into k classes in a frequency
table, where xi is the midpoint and fi is the frequency of the ith class interval. If x
is the mean for the grouped data, then the sample standard deviation s for the
grouped data is
a 1xi - x2 fi
k
2
s = i=1 (5)
H
where n = g i = 1 fi = total number of measurements. If x1, x2,c, xn is the whole
n - 1
k
population with mean m, then the population standard deviation S is given by
a 1xi - m2 fi
n
2
i=1
s =
H n
*
In this section, we restrict our interest to the sample standard deviation.
EXAMPLE 2 Finding the Standard Deviation for Grouped Data Find the standard deviation
for each set of grouped sample data.
4 4 4
2 2
1 1 1 1
0
8 9 10 11 12 8 9 10 11 12
Mean 5 10 Mean 5 10
SOLUTION
18 - 102 2 112 + 19 - 102 2 122 + 110 - 102 2 142 + 111 - 102 2 122 + 112 - 102 2 112
(A) s =
A 10 - 1
12
= ≈ 1.15
A9
18 - 102 2 142 + 19 - 102 2 112 + 110 - 102 2 102 + 111 - 102 2 112 + 112 - 102 2 142
(B) s =
A 10 - 1
34
= ≈ 1.94
A9
Comparing the results of parts (A) and (B) in Example 2, we find that the larger
standard deviation is associated with the data that deviate furthest from the mean.
Matched Problem 2 Find the standard deviation for the grouped sample data
shown below.
2 2 2 2 2
8 9 10 11 12
Mean 5 10
Remark—Figure 3 illustrates the shortcut computation of the mean and standard

deviation on a graphing calculator when the data are grouped. The sample data of
Example 2A are entered. List L 1 contains the midpoints of the class intervals, and
list L 2 contains the corresponding frequencies. The mean, standard deviation, and
other one-variable statistics are then calculated immediately.
Figure 3
(A) (B)
Significance of Standard Deviation

The standard deviation can give us additional information about a frequency distribution
of a set of raw data. Suppose we draw a smooth curve through the midpoints of the tops of
the rectangles forming a histogram for a fairly large frequency distribution (see Fig. 4). If
the resulting curve is approximately bell shaped, then it can be shown that approximately
68% of the data will lie in the interval from x - s to x + s, about 95% of the data will lie
in the interval from x - 2s to x + 2s, and almost all the data will lie in the interval from
x - 3s to x + 3s. We will have much more to say about this in Section 9.5.
68%
x 2 3s x 2 2s x2s x x1s x 1 2s x 1 3s
Figure 4
Exercises 9.3
Skills Warm-up Exercises
2. a 1xi - 102 2
5
W In Problems 1–8, find the indicated sum. (If necessary, review if x1 = 8, x2 = 9, x3 = 10, x4 = 11, x5 = 12
Section C.1). i=1
1. a 1xi - 22 2 3. a 1xi + 32 2
3 4
if x1 = - 1, x2 = 1, x3 = 6 if x1 = - 4, x2 = - 3, x3 = - 2, x4 = - 3
i=1 i=1
4. a 1xi + 152 2
3
if 12. Interval Frequency
i=1 0.5–3.5 5
x1 = - 20, x2 = - 15, x3 = - 10 3.5–6.5 1
5. a 1xi - 42 2 fi
3 6.5–9.5 2
if 9.5–12.5 7
i=1
x1 = 0, x2 = 4, x3 = 8, f1 = 1, f2 = 1, f3 = 1
6. a 1xi - 42 2 fi
3 statement is always true, explain why. If not, give a counterexample.
if
i=1 13. The range for a set of sample measurements is less than or
x1 = 0, x2 = 4, x3 = 8, f1 = 2, f2 = 1, f3 = 2 equal to the range of the whole population.
7. a 1xi - 152 2 fi
5 14. The range for a set of measurements is greater than or equal to 0.
if
i=1 15. The sample standard deviation is less than or equal to the
xi = 5i, fi = 2, for i = 1, 2, c, 5 sample variance.
8. a 1xi - 3.52 2 fi
4 16. Given a set of sample measurements that are not all equal, the
if sample standard deviation is a positive real number.
i=1
x i = i + 1, f i = 3, for i = 1, 2, 3, 4 17. If x1, x2 is the whole population, then the population standard
deviation is equal to one-half the distance from x1 to x2.
A 9. (A) Find the mean and standard deviation of the following 18. The sample variance of a set of measurements is always less
set of ungrouped sample data. than the population variance.
4 2 3 5 3 1 6 4 2 3 C 19. A data set is formed by recording the sums in 100 rolls of
(B) What proportion of the measurements lies within 1 a pair of dice. A second data set is formed by recording the
standard deviation of the mean? Within 2 standard results of 100 draws of a ball from a box containing 11 balls
deviations? Within 3 standard deviations? numbered 2 through 12.
(C) Based on your answers to part (B), would you conjec- (A) Which of the two data sets would you expect to have the
ture that the histogram is approximately bell shaped? smaller standard deviation? Explain.
Explain. (B) To obtain evidence for your answer to part (A), use a
(D) To confirm your conjecture, construct a histogram with graphing calculator to simulate both experiments and
class interval width 1, starting at 0.5. compute the standard deviations of each data set.
10. (A) Find the mean and standard deviation of the following 20. A data set is formed by recording the results of rolling a fair die
set of ungrouped sample data. 200 times. A second data set is formed by rolling a pair of dice
200 times, each time recording the minimum of the two numbers.
3 5 1 2 1 5 4 5 1 3
(A) Which of the two data sets would you expect to have the
(B) What proportion of the measurements lies within 1
smaller standard deviation? Explain.
standard deviation of the mean? Within 2 standard devia-
tions? Within 3 standard deviations? (B) To obtain evidence for your answer to part (A), use a
graphing calculator to simulate both experiments and
(C) Based on your answers to part (B), would you conjec-
compute the standard deviations of each data set.
ture that the histogram is approximately bell shaped?
Explain.
(D) To confirm your conjecture, construct a histogram with
class interval width 1, starting at 0.5.
Applications
Find the mean and standard deviation for each of the sample data
B In Problems 11 and 12, find the standard deviation for each set of
grouped sample data using formula (5) on page 503. sets given in Problems 21–28.
Use the suggestions in the remarks following Examples 1 and 2 to
11. Interval Frequency perform some of the computations.
0.5–3.5 2 21. Earnings per share. The earnings per share (in dollars)
3.5–6.5 5 for 12 companies selected at random from the list of Fortune
6.5–9.5 7 500 companies are:
9.5–12.5 1
2.35 1.42 8.05 6.71
3.11 2.56 0.72 4.17
5.33 7.74 3.88 6.21
22. Checkout times. The checkout times (in minutes) for 12 26. Nutrition: animals. The mouse weights (in grams) of a
randomly selected customers at a large supermarket during random sample of 100 mice involved in a nutrition experi-
the store’s busiest time are: ment are:
Interval Frequency
4.6 8.5 6.1 7.8
10.9 9.3 11.4 5.8 41.5–43.5 3
9.7 8.8 6.7 13.2 43.5–45.5 7
45.5–47.5 13
23. Quality control. The lives (in hours of continuous use) of 47.5–49.5 17
100 randomly selected flashlight batteries are: 49.5–51.5 19
51.5–53.5 17
Interval Frequency 53.5–55.5 15
6.95–7.45 2 55.5–57.5 7
7.45–7.95 10 57.5–59.5 2
7.95–8.45 23
27. Reading scores. The grade-level reading scores from a read-
8.45–8.95 30
ing test given to a random sample of 12 students in an urban
8.95–9.45 21 high school graduating class are:
9.45–9.95 13
9 11 11 15
9.95–10.45 1
10 12 12 13
24. Stock analysis. The price–earnings ratios of 100 randomly 8 7 13 12
selected stocks from the New York Stock Exchange are: 28. Grade-point average. The grade-point averages of a ran-
dom sample of 100 students from a university’s graduating
Interval Frequency class are:
-0.5–4.5 5
Interval Frequency
4.5–9.5 54
9.5–14.5 25 1.95–2.15 21
14.5–19.5 13 2.15–2.35 19
19.5–24.5 0 2.35–2.55 17
24.5–29.5 1 2.55–2.75 14
29.5–34.5 2 2.75–2.95 9
2.95–3.15 6
25. Medicine. The reaction times (in minutes) of a drug given to 3.15–3.35 5
a random sample of 12 patients are: 3.35–3.55 4
3.55–3.75 3
4.9 5.1 3.9 4.2 3.75–3.95 2
6.4 3.4 5.8 6.1
5.0 5.6 5.8 4.6 Answers to Matched Problems
1. s ≈ 0.19
2. s ≈ 1.49 (a value between those found in Example 2, as
expected)
9.4 Bernoulli Trials and Binomial Distributions

■■ Bernoulli Trials In Section 9.1, we discussed frequency and relative frequency distributions, which
■■ Binomial Formula: Brief Review
were represented by tables and histograms. Frequency distributions and their cor-
responding probability distributions based on actual observations are empirical in
■■ Binomial Distribution nature. But there are many situations in which it is of interest to determine the kind of
■■ Application relative frequency distribution we might expect before any data have been collected.
What we have in mind is a theoretical, or hypothetical, probability distribution—
that is, a probability distribution based on assumptions and theory rather than actual
observations or measurements. Theoretical probability distributions are used to
approximate properties of real-world distributions, assuming that the theoretical and
empirical distributions are closely matched.
Of the many interesting theoretical probability distributions, one in particular is
the binomial distribution. The reason for the name “binomial distribution” is that the
distribution is closely related to the binomial expansion of 1q + p2 n, where n is a
SECTION 9.4 Bernoulli Trials and Binomial Distributions 507
natural number. We start the discussion with a particular type of experiment called a
Bernoulli experiment, or trial.
Bernoulli Trials
If we toss a coin, either a head occurs or it does not. If we roll a die, either a 3 shows
or it fails to show. If you are vaccinated for smallpox, either you contract smallpox or
you do not. What do all these situations have in common? All can be classified as ex-
periments with two possible outcomes, each the complement of the other. An experi-
ment for which there are only two possible outcomes, E or E′, is called a Bernoulli
experiment, or trial, named after Jacob Bernoulli (1654–1705), the Swiss scientist
and mathematician who was one of the first to study the probability problems related
to a two-outcome experiment.
In a Bernoulli experiment or trial, it is customary to refer to one of the two out-
comes as a success S and to the other as a failure F. If we designate the probability
Reminder of success by
Uppercase “P” is an abbreviation for P1S2 = p
the word “probability.” Lowercase
“p” stands for a number between 0
then the probability of failure is
and 1, inclusive. P1F2 = 1 - p = q Note: p + q = 1
EXAMPLE 1 Probability of Success in a Bernoulli Trial Suppose that we roll a fair die and
ask for the probability of a 6 turning up. This can be viewed as a Bernoulli trial by
identifying success with a 6 turning up and failure with any of the other numbers
turning up. So,
p = 16 and q = 1 - 16 = 56
Matched Problem 1 Find p and q for a single roll of a fair die, where success
is a number divisible by 3 turning up.
Now, suppose that a Bernoulli trial is repeated a number of times. We might try
to determine the probability of a given number of successes out of the given number
of trials. For example, we might be interested in the probability of obtaining exactly
three 5’s in six rolls of a fair die or the probability that 8 people will not catch influ-
enza out of the 10 who have been inoculated.
Suppose that a Bernoulli trial is repeated five times so that each trial is com-
pletely independent of any other, and p is the probability of success on each trial.
Then the probability of the outcome SSFFS would be
P1SSFFS2 = P1S2P1S2P1F2P1F2P1S2 See Section 7.3.

= ppqqp
= p3q2
In general, we define a sequence of Bernoulli trials as follows:
DEFINITION Bernoulli Trials

A sequence of experiments is called a sequence of Bernoulli trials, or a binomial
experiment, if
1. Only two outcomes are possible in each trial.
2. The probability of success p for each trial is a constant (probability of failure
is then q = 1 - p).
3. All trials are independent.
EXAMPLE 2 Probability of an Outcome of a Binomial Experiment If we roll a fair die five

times and identify a success in a single roll with a 1 turning up, what is the prob-
ability of the sequence SFFSS occurring?
SOLUTION
1 5
p = 6 q = 1 - p = 6
P1SFFSS2 = pqqpp
= p3q2
= 1 16 23 1 56 22 ≈ .003
Matched Problem 2 In Example 2, find the probability of the outcome FSSSF.
If we roll a fair die five times, what is the probability of obtaining exactly three
1’s? Notice how this problem differs from Example 2. In that example we looked at
only one way that three 1’s can occur. Then in Matched Problem 2 we saw another
way. So exactly three 1’s may occur in the following two sequences (among others):
SFFSS FSSSF
We found that the probability of each sequence occurring is the same, namely,
1 16 23 1 56 22
How many more sequences will produce exactly three 1’s? To answer this ques-
tion, think of the number of ways that the following five blank positions can be filled
with three S’s and two F’s:
b1 b2 b3 b4 b5
A given sequence is determined, once the S’s are assigned. We are interested in the
number of ways three blank positions can be selected for the S’s out of the five avail-
able blank positions b1, b2, b3, b4, and b5. This problem should sound familiar—it is
the problem of finding the number of combinations of 5 objects taken 3 at a time, that
is, 5C3. So, the number of different sequences of successes and failures that produce
exactly three successes (exactly three 1’s) is
5!
5C3 = = 10
3!2!
Since the probability of each sequence is the same,
p3q2 = 1 16 23 1 56 22
and there are 10 mutually exclusive sequences that produce exactly three 1’s,
1 3 5 2
P1exactly three successes2 = 5C3 a b a b
6 6
5! 1 3 5 2
= a b a b
3!2! 6 6
1 3 5 2
= 1102 a b a b ≈ .032
6 6
Reasoning in essentially the same way, the following important theorem can be
proved:
THEOREM 1 Probability of x Successes in n Bernoulli Trials

The probability of exactly x successes in n independent repeated Bernoulli trials,
with the probability of success of each trial p (and of failure q), is
P1x successes2 = nCx pxqn - x (1)
EXAMPLE 3 Probability of x Successes in n Bernoulli Trials If a fair die is rolled five times,
what is the probability of rolling
(A) Exactly two 3’s?
(B) At least two 3’s?
SOLUTION
(A) Use formula (1) with n = 5, x = 2, and p = 16:
1 2 5 3
P1x = 22 = 5C2 a b a b
6 6
5! 1 2 5 3
= a b a b ≈ .161
2!3! 6 6
(B) Notice how this problem differs from part (A). Here we have
P1x Ú 22 = P1x = 22 + P1x = 32 + P1x = 42 + P1x = 52
It is actually easier to compute the probability of the complement of this event,
P1x 6 22, and use
P1x Ú 22 = 1 - P1x 6 22
where
P1x 6 22 = P1x = 02 + P1x = 12
We now compute P1x = 02 and P1x = 12:
1 0 5 5 1 1 5 4
P1x = 02 = 5C0 a b a b P1x = 12 = 5C1 a b a b
6 6 6 6
5 5 5! 1 1 5 4
= a b ≈ .402 = a b a b ≈ .402
6 1!4! 6 6
Therefore,
P1x 6 22 = .402 + .402 = .804
and
P1x Ú 22 = 1 - .804 = .196
Matched Problem 3 Using the same die experiment as in Example 3, what is
the probability of rolling
(A) Exactly one 3? (B) At least one 3?
Binomial Formula: Brief Review

Before extending Bernoulli trials to binomial distributions, it is worthwhile to review
the binomial formula, which is discussed in detail in Appendix C.3. To start, let us
calculate directly the first five natural number powers of 1a + b2 n:
1a + b2 1 = a + b
1a + b2 2 = a2 + 2ab + b2
1a + b2 3 = a3 + 3a2b + 3ab2 + b3
1a + b2 4 = a4 + 4a3b + 6a2b2 + 4ab3 + b4
1a + b2 5 = a5 + 5a4b + 10a3b2 + 10a2b3 + 5ab4 + b5
In general, it can be shown that a binomial expansion is given by the well-known
binomial formula:
Binomial Formula
For n a natural number,
1a + b2 n = n C 0 an + nC1 an - 1b + nC2 an - 2b2 + g + nCn bn
EXAMPLE 4 Finding Binomial Expansions Use the binomial formula to expand 1q + p2 3.

SOLUTION
1q + p23 = 3C0q3 + 3C1q2p + 3C2qp2 + 3C3p3

= q3 + 3q2p + 3qp2 + p3
Matched Problem 4 Use the binomial formula to expand 1q + p2 4.
Binomial Distribution
We now generalize the discussion of Bernoulli trials to binomial distributions. We
start by considering a sequence of three Bernoulli trials. Let the random variable X3
represent the number of successes in three trials, 0, 1, 2, or 3. We are interested in the
probability distribution for this random variable.
Which outcomes of an experiment consisting of a sequence of three Bernoulli
trials lead to the random variable values 0, 1, 2, and 3, and what are the probabilities
associated with these values? Table 1 answers these questions.
Table 1
Probability of X3
Simple Event Simple Event x successes in 3 trials P1X3 = x2
FFF qqq = q3 0 q3
FFS 2
qqp = q p 1 3q2p
FSF 2
qpq = q p
SFF pqq = q2p
FSS qpp = qp2 2 3qp2
SFS pqp = qp2
SSF ppq = qp2
SSS ppp = p3 3 p3
The terms in the last column of Table 1 are the terms in the binomial expansion
of 1q + p2 3, as we saw in Example 4. The last two columns in Table 1 provide a
probability distribution for the random variable X3. Note that both conditions for a
probability distribution (see Section 7.5) are met:
1. 0 … P1X3 = x2 … 1, x ∊ 50, 1, 2, 36
2. 1 = 13 = 1q + p2 3 Recall that q + p = 1.
3 2 2 3
= 3C0q + 3C1q p + 3C2qp + 3C3p
= q3 + 3q2p + 3qp2 + p3
= P1X3 = 02 + P1X3 = 12 + P1X3 = 22 + P1X3 = 32
Reasoning in the same way for the general case, we see why the probability dis-
tribution of a random variable associated with the number of successes in a sequence
of n Bernoulli trials is called a binomial distribution—the probability of each num-
ber is a term in the binomial expansion of 1q + p2 n. For this reason, a sequence of
Bernoulli trials is often referred to as a binomial experiment. In terms of a formula,
which we already discussed from another point of view (see Theorem 1), we have
DEFINITION Binomial Distribution

P1Xn = x2 = P1x successes in n trials2
= nCx pxq n - x x ∊ 50, 1, 2,c, n6
where p is the probability of success and q is the probability of failure on each trial.
Informally, we will write P1x2 in place of P1Xn = x2.
EXAMPLE 5 Constructing Tables and Histograms for Binomial Distributions Suppose a fair
die is rolled three times and success on a single roll is considered to be rolling a
number divisible by 3. For the binomial distribution,
(A) Write the probability function.
(B) Construct a table.
(C) Draw a histogram.
SOLUTION
(A) p = 13 Since two numbers out of six are divisible by 3
2
q = 1 - p = 3
n = 3
Therefore,
P1x2 = P1x successes in 3 trials2 = 3C x 1 13 2x 1 23 23 - x
(B) (C) P(x)
x P1x2
.5
.44
0 1 21 2
1 0 2 3
3C0 3 3 ≈ .30
.4
1 C 1 21 2
1 1 2 2
3 1 3 3 ≈ .44
.30
2 1 21 2
1 2 2 1
3C2 3 3 ≈ .22 .3
.22
3C3 1 3 2 1 3 2
3 1 3 2 0
≈ .04 .2
1.00
.1
.04
0
0 1 2 3
Number of successes, x
Figure 1
If we actually performed the binomial experiment in Example 5 a large number

of times with a fair die, we would find that we would roll no number divisible by 3
in three rolls of a die about 30% of the time, one number divisible by 3 in three rolls
about 44% of the time, two numbers divisible by 3 in three rolls about 22% of the
time, and three numbers divisible by 3 in three rolls only 4% of the time. Note that
the sum of all the probabilities is 1, as it should be.
The graphing calculator command in Figure 2A simulates 100 repetitions of the
binomial experiment in Example 5. The number of successes on each trial is stored
in list L 1. From Figure 2B, which shows a histogram of L 1, we note that the em-
48
pirical probability of rolling one number divisible by 3 in three rolls is 100 = 48%,
close to the theoretical probability of 44%. The empirical probabilities of 0, 2, or 3
successes also would be close to the corresponding theoretical probabilities.
(A) (B)
Figure 2
Matched Problem 5 Repeat Example 5, where the binomial experiment con-

sists of two rolls of a die instead of three rolls.
Let X be a random variable with probability distribution

xi x1 x2 c xn
pi p1 p2 c pn
In Section 7.5 we defined the expected value of X to be
E1X2 = x1 p1 + x2 p2 + g + xn pn
The expected value of X is also called the mean of the random variable X, often denoted
by m. The standard deviation of a random variable X having mean m is defined by
s = 21x1 - m2 2 # p1 + 1x2 - m2 2 # p2 + g + 1xn - m2 2 # pn
If a random variable has a binomial distribution, where n is the number of Bernoulli
trials, p is the probability of success, and q the probability of failure, then the mean
and standard deviation are given by the following formulas:
Mean: m = np
Standard deviation: s = 2npq
CONCEPTUAL INSIGHT
Let the random variable X3 denote the number x of successes in a sequence of
three Bernoulli trials. Then x = 0, 1, 2, or 3. The expected value of X3 is given by
E1X3 2 = 0 # P102 + 1 # P112 + 2 # P122 + 3 # P132 See Table 1.
= 0 + 1 # 3q2p + 2 # 3qp2 + 3 # p3 Factor out 3p.
2 2
= 3p1q + 2qp + p 2 Factor the perfect square.
= 3p1q + p2 2 q + p = 1
= 3p
This proves the formula m = np in the case n = 3. Similar but more complicated
computations can be used to justify the general formulas m = np and s = 1npq for
the mean and standard deviation of random variables having binomial distributions.
EXAMPLE 6 Computing the Mean and Standard Deviation of a Binomial Distribution Com-
pute the mean and standard deviation for the random variable in Example 5.
1 1 2
SOLUTION n = 3 p = 3 q = 1 - 3 = 3
m = np = 3 1 13 2 = 1 s = 1npq = 23 1 13 2 1 23 2 ≈ .82
Matched Problem 6 Compute the mean and standard deviation for the ran-
dom variable in Matched Problem 5.
Application
Binomial experiments are associated with a wide variety of practical problems: in-
dustrial sampling, drug testing, genetics, epidemics, medical diagnosis, opinion polls,
analysis of social phenomena, qualifying tests, and so on. We will now consider one
application in detail.
EXAMPLE 7 Patient Recovery The probability of recovering after a particular type of opera-
tion is .5. Eight patients undergo this operation. For the binomial distribution,
(C) Construct a histogram.
(D) Find the mean and standard deviation.
SOLUTION
(A) Letting a recovery be a success, we have
p = .5 q = 1 - p = .5 n = 8
Hence,
P1x2 = P1exactly x successes in 8 trials2 = 8C x 1.52x 1.528 - x = 8C x 1.528
(B) x P1x2
(C) P(x)
0 8C0 1.52
8
≈ .004 .3
.273
1 8
8C1 1.52 ≈ .031
2 8C2 1.52
8
≈ .109 .219 .219
3 8
≈ .219 .2
8C3 1.52
4 8C4 1.52
8
≈ .273
5 8
≈ .219 .109 .109
8C5 1.52
.1
6 8C6 1.52
8
≈ .109
7 8C7 1.52
8
≈ .031 .031 .031
.004 .004
8 8C8 1.52
8
≈ .004 0
0 1 2 3 4 5 6 7 8
.999 ≈ 1
The discrepancy in the sum is due to
round-off errors.
(D) m = np = 81.52 = 4 s = 1npq = 181.521.52 ≈ 1.41
Matched Problem 7 Repeat Example 7 for four patients.
Exercises 9.4
A Evaluate nCx pxqn - x for the values of n, x, and p given in Prob- B In Problems 19–24, round answers to four decimal places. A fair
lems 1–6. die is rolled four times. Find the probability of obtaining
1 1
1. n = 5, x = 1, p = 2 2. n = 5, x = 2, p = 2 19. A 6, 6, 5, and 5 in that order.
3. n = 6, x = 3, p = .4 4. n = 6, x = 6, p = .4 20. Two 6’s and two 5’s in any order.
2 1
5. n = 4, x = 3, p = 3 6. n = 4, x = 3, p = 3 21. Exactly two 6’s. 22. Exactly three 6’s.
In Problems 7–12, a fair coin is tossed four times. What is the 23. No 6’s. 24. At least two 6’s.
probability of obtaining 25. If a baseball player has a batting average of .350, what is the
7. A head on the first toss and tails on each of the other tosses? probability that the player will get the following number of
hits in the next four times at bat?
8. Exactly one head?
(A) Exactly 2 hits (B) At least 2 hits
9. At least three tails? 26. If a true–false test with 10 questions is given, what is the
10. Tails on each of the first three tosses? probability of scoring
(A) Exactly 70% just by guessing?
11. No heads?
(B) 70% or better just by guessing?
12. Four heads?
27. A multiple-choice test consists of 10 questions, each with
In Problems 13–18, construct a histogram for the binomial distri- choices A, B, C, D, E (exactly one choice is correct). Which
bution P1x2 = nC x p xqn - x, and compute the mean and standard is more likely if you simply guess at each question: all your
deviation if answers are wrong, or at least half are right? Explain.
1 3
13. n = 3, p = 4 14. n = 3, p = 4 28. If 60% of the electorate supports the mayor, what is the prob-
15. n = 4, p = 1
16. n = 5, p = 1 ability that in a random sample of 10 voters, fewer than half
3 3
support her?
17. n = 5, p = 0 18. n = 4, p = 1
Construct a histogram for each of the binomial distributions in (B) Use a graphing calculator to simulate 100 repetitions
Problems 29–32. Compute the mean and standard deviation for of the binomial experiment, and compare the mean and
each distribution. standard deviation of the numbers of 7’s or 11’s from the
simulation to the answers for part (A).
29. P1x2 = 6C x 1.42x 1.626 - x 30. P1x2 = 6C x 1.62x 1.426 - x
31. P1x2 = 8C x 1.32x 1.728 - x 32. P1x2 = 8C x 1.72x 1.328 - x
Applications
In Problems 33 and 34, use a graphing calculator to construct a
probability distribution table. 43. Management training. Each year a company selects a
33. A random variable represents the number of successes in number of employees for a management training program
20 Bernoulli trials, each with probability of success p = .85. at a university. On average, 70% of those sent complete the
program. Out of 7 people sent, what is the probability that
(A) Find the mean and standard deviation of the random
variable. (A) Exactly 5 complete the program?
(B) Find the probability that the number of successes lies (B) 5 or more complete the program?
within 1 standard deviation of the mean.
44. Employee turnover. If the probability of a new employee in a
34. A random variable represents the number of successes in fast-food chain still being with the company at the end of 1 year
20 Bernoulli trials, each with probability of success p = .45. is .6, what is the probability that out of 8 newly hired people,
(A) 5 will still be with the company after 1 year?
variable.
(B) 5 or more will still be with the company after 1 year?
(B) Find the probability that the number of successes lies
within 1 standard deviation of the mean. 45. Quality control. A manufacturing process produces, on
average, 6 defective items out of 100. To control quality, each
C In Problems 35 and 36, a coin is loaded so that the probability
day a sample of 10 completed items is selected at random
of a head occurring on a single toss is 34. In five tosses of the coin,
and inspected. If the sample produces more than 2 defec-
what is the probability of getting
tive items, then the whole day’s output is inspected, and the
35. All heads or all tails? manufacturing process is reviewed. What is the probability of
36. Exactly 2 heads or exactly 2 tails? this happening, assuming that the process is still producing
6% defective items?
37. Toss a coin three times or toss three coins simultaneously,
and record the number of heads. Repeat the binomial experi- 46. Guarantees. A manufacturing process produces, on aver-
ment 100 times and compare your relative frequency distribu- age, 3% defective items. The company ships 10 items in each
tion with the theoretical probability distribution. box and wants to guarantee no more than 1 defective item per
box. If this guarantee applies to each box, what is the prob-
38. Roll a die three times or roll three dice simultaneously, and ability that the box will fail to meet the guarantee?
record the number of 5’s that occur. Repeat the binomial
experiment 100 times and compare your relative frequency 47. Quality control. A manufacturing process produces, on
distribution with the theoretical probability distribution. average, 5 defective items out of 100. To control quality, each
day a random sample of 6 completed items is selected and
39. Find conditions on p that guarantee the histogram for a bino- inspected. If success on a single trial (inspection of 1 item)
mial distribution is symmetrical about x = n>2. Justify your is finding the item defective, then the inspection of each of
answer. 6 items in the sample constitutes a binomial experiment. For
40. Consider two binomial distributions for 1,000 repeated Ber- the binomial distribution,
noulli trials—the first for trials with p = .15, and the second (A) Write the probability function.
for trials with p = .85. How are the histograms for the two
distributions related? Explain. (B) Construct a table.
41. A random variable represents the number of heads in ten (C) Draw a histogram.
tosses of a coin. (D) Compute the mean and standard deviation.
48. Management training. Each year a company selects 5 em-
variable.
ployees for a management training program at a university.
(B) Use a graphing calculator to simulate 200 repetitions On average, 40% of those sent complete the course in the top
of the binomial experiment, and compare the mean and 10% of their class. If we consider an employee finishing in
standard deviation of the numbers of heads from the the top 10% of the class a success in a binomial experiment,
simulation to the answers for part (A). then for the 5 employees entering the program, there exists a
42. A random variable represents the number of times a sum of binomial distribution involving P(x successes out of 5). For
7 or 11 comes up in ten rolls of a pair of dice. the binomial distribution,
(A) Find the mean and standard deviation of the random (A) Write the probability function.
variable.
(C) Draw a histogram. (C) Draw a histogram.

(D) Compute the mean and standard deviation. (D) Compute the mean and standard deviation.
49. Medical diagnosis. A tuberculosis patient is given a chest 58. Sociology. The probability that a marriage will end in divorce
x-ray. Four tuberculosis x-ray specialists examine each x-ray within 10 years is .35. What are the mean and standard devia-
independently. If each specialist can detect tuberculosis 80% tion for the binomial distribution involving 1,000 marriages?
of the time when it is present, what is the probability that
59. Sociology. If the probability is .55 that a marriage will end
at least 1 of the specialists will detect tuberculosis in this
in divorce within 20 years, what is the probability that out of
patient?
6 couples just married, in the next 20 years
50. Harmful drug side effects. A pharmaceutical laboratory
(A) None will be divorced?
claims that a drug causes serious side effects in 20 people out of
1,000, on average. To check this claim, a hospital administers (B) All will be divorced?
the drug to 10 randomly chosen patients and finds that 3 suffer
(C) Exactly 2 will be divorced?
from serious side effects. If the laboratory’s claims are correct,
what is the probability that the hospital gets these results? (D) At least 2 will be divorced?
51. Genetics. The probability that brown-eyed parents, both Answers to Matched Problems
with the recessive gene for blue eyes, will have a child with
brown eyes is .75. If such parents have 5 children, what is the 1. p = 13, q = 2
3 2. p3q2 = 1 16 23 1 56 22 ≈ .003
probability that they will have
3. (A) .402
(A) All blue-eyed children? (B) 1 - P1x = 02 = 1 - .402 = .598
(B) Exactly 3 children with brown eyes? 4. 4
+ 4C1q3p + 4C2q2p2 + 4C3qp3 + 4C4p4 =
4C0 q
(C) At least 3 children with brown eyes? q4 + 4q3p + 6q2p2 + 4qp3 + p4

5. (A) P1x2 = P1x successes in 2 trials2
52. Gene mutations. The probability of gene mutation under a
given level of radiation is 3 * 10-5. What is the probability
1
2C x 3 1 2 x 1 23 2 2 - x, x ∊ 50, 1, 26
of at least 1 gene mutation if 105 genes are exposed to this (B) x P1x2 (C) P(x)
level of radiation? 0 4 .5 .44 .44
9 ≈ .44
.4
53. Epidemics. If the probability of a person contracting influ- 1 4
≈ .44 .3
9
enza on exposure is .6, consider the binomial distribution for 2 1 .2
9 ≈ .11 .11
a family of 6 that has been exposed. For this distribution, .1
0
(A) Write the probability function. 0 1 2
(C) Draw a histogram. 6. m ≈ .67; s ≈ .67
7. (A) P1x2 = P1exactly x successes in 4 trials2 = 4Cx 1.52 4
(D) Compute the mean and standard deviation. (B) x P1x2
54. Drug side effects. The probability that a given drug will 0 .06
produce a serious side effect in a person using the drug is .02. 1 .25
In the binomial distribution for 450 people using the drug, 2 .38
what are the mean and standard deviation? 3 .25
55. Testing. A multiple-choice test is given with 5 choices (only 4 .06
one is correct) for each of 10 questions. What is the prob- 1.00
ability of passing the test with a grade of 70% or better just
(C) P(x)
by guessing?
.4 .38
56. Opinion polls. An opinion poll based on a small sample can .3 .25 .25
be unrepresentative of the population. To see why, assume .2
that 40% of the electorate favors a certain candidate. If a .1 .06 .06
random sample of 7 is asked their preference, what is the 0
probability that a majority will favor this candidate? 0 1 2 3 4
57. Testing. A multiple-choice test is given with 5 choices (only
one is correct) for each of 5 questions. Answering each of the (D) m = 2; s = 1
5 questions by guessing constitutes a binomial experiment
with an associated binomial distribution. For this distribution,

■■ Normal Distribution Normal Distribution
■■ Areas under Normal Curves If we take the histogram for a binomial distribution, such as the one we drew for
■■ Approximating a Binomial Example 7, Section 9.4 1n = 8, p = .52, and join the midpoints of the tops of the
Distribution with a Normal bars with a smooth curve, we obtain the bell-shaped curve in Figure 1.
Distribution P(x)
.3
.2
.1
0
0 1 2 3 4 5 6 7 8
Figure 1 Binomial distribution and bell-shaped curve
The mathematical foundation for this type of curve was established by Abraham
De Moivre (1667–1754), Pierre Laplace (1749–1827), and Carl Gauss (1777–1855).
The bell-shaped curves studied by these famous mathematicians are called normal
curves or normal probability distributions, and their equations are determined by
the mean M and standard deviation S of the distribution. Figure 2 illustrates three
normal curves with different means and standard deviations.
CONCEPTUAL INSIGHT
The equation for a normal curve is fairly complicated:
1 2 2
f1x2 = e-1x - m2 >2s
s 22p
where p ≈ 3.1416 and e ≈ 2.7183. Given the values of m and s, however,
the function is completely specified, and we could plot points or use a graphing
calculator to produce its graph. Substituting x + h for x produces an equation of
the same form but with a different value of m. Therefore, in the terminology of
Section 1.2, any horizontal translation of a normal curve is another normal curve.
m 5 20
s51
m 5 15 m 5 50
s53 s54
x
0 10 20 30 40 50 60 70
m
Figure 2 Normal probability distributions
SECTION 9.5 Normal Distributions 517
Until now we have dealt with discrete random variables, that is, random vari-
ables that assume a finite or “countably infinite” number of values (we have dealt
only with the finite case). Random variables associated with normal distributions
are continuous in nature; they assume all values over an interval on a real number
line. These are called continuous random variables. Random variables associated
with people’s heights, lightbulb lifetimes, or the lengths of time between breakdowns
of a copy machine are continuous. The following is a list of some of the important
properties of normal curves (normal probability distributions of a continuous random
variable):
Properties of Normal Curves

1. Normal curves are bell shaped and symmetrical with respect to a vertical line.
2. The mean is the real number at the point where the axis of symmetry intersects
the horizontal axis.
3. The shape of a normal curve is completely determined by its mean and a positive
real number called the standard deviation. A small standard deviation indicates a
tight clustering about the mean and a tall, narrow curve; a large standard devia-
tion indicates a large deviation from the mean and a broad, flat curve (see Fig. 2).
4. Irrespective of the shape, the area between the curve and the x axis is always 1.
5. Irrespective of the shape, 68.3% of the area will lie within an interval of 1 stan-
dard deviation on either side of the mean, 95.4% within 2 standard deviations on
either side, and 99.7% within 3 standard deviations on either side (see Fig. 3).
99.7%
95.4%
68.3%
m 2 3s m 2 2s m 2 1s m m 1 1s m 1 2s m 1 3s x
23 22 21 0 1 2 3 z
Figure 3 Normal curve areas
The normal probability distribution is the most important of all theoretical distri-
butions, and it is a useful tool for solving practical problems. Not only does a normal
curve provide a good approximation for a binomial distribution for large n, but it
also approximates many other relative frequency distributions. For example, normal
curves often provide good approximations for the relative frequency distributions for
heights and weights of people, measurements of manufactured parts, scores on IQ
tests, college entrance examinations, civil service tests, and measurements of errors
in laboratory experiments.
Areas under Normal Curves

To use normal curves in practical problems, we must be able to determine areas under
different parts of a normal curve. Remarkably, the area under a normal curve between
a mean m and a given number of standard deviations to the right (or left) of m is
the same, regardless of the shape of the normal curve. For example, the area under the
normal curve with m = 3, s = 5 from m = 3 to m + 1.5s = 10.5 is equal to
the area under the normal curve with m = 15, s = 2 from m = 15 to m + 1.5s = 18
(see Fig. 4, noting that the shaded regions have the same areas, or equivalently, the
same numbers of pixels). Therefore, such areas for any normal curve can be easily
determined from the areas for the standard normal curve, that is, the normal curve
with mean 0 and standard deviation 1. In fact, if z represents the number of standard
deviations that a measurement x is from a mean m, then the area under a normal curve
from m to m + zs equals the area under the standard normal curve from 0 to z (see
Fig. 5). Appendix D lists those areas for the standard normal curve.
f (x)
Area
corresponding
to z
m m 1 zs x
0 z z
Figure 4 Figure 5 Areas and z values
EXAMPLE 1 Finding Probabilities for a Normal Distribution A manufacturing process pro-

duces lightbulbs with life expectancies that are normally distributed with a mean of
500 hours and a standard deviation of 100 hours. What percentage of the lightbulbs
can be expected to last between 500 and 670 hours?
SOLUTION To answer this question, we determine how many standard deviations
670 is from 500, the mean. This is done by dividing the distance between 500 and
670 by 100, the standard deviation. Thus,
670 - 500 170

z = = = 1.70
100 100
That is, 670 is 1.7 standard deviations from 500, the mean. Referring to Appendix D,
we see that .4554 corresponds to z = 1.70. And since the total area under a normal
curve is 1, we conclude that 45.54% of the lightbulbs produced will last between 500
and 670 hours (see Fig. 6).
f (x)
.4554
or
45.54%
500 670 x
0 1.7 z
Figure 6 Lightbulb life expectancy: positive z
Matched Problem 1 What percentage of the lightbulbs in Example 1 can be

expected to last between 500 and 750 hours?
In general, to find how many standard deviations that a measurement x is from a

mean m, first determine the distance between x and m and then divide by s:
distance between x and m x - m
z = =
standard deviation s
EXAMPLE 2 Finding Probabilities for a Normal Distribution From all lightbulbs produced
(see Example 1), what is the probability that a lightbulb chosen at random lasts
between 380 and 500 hours?
SOLUTION To answer this, we first find z:
x - m 380 - 500
z = = = - 1.20
s 100
It is usually a good idea to draw a rough sketch of a normal curve and insert relevant
data (see Fig. 7).
f (x)
.3849
or
38.49%
380 500 x
21.20 0 z
Figure 7 Lightbulb life expectancy: negative z
Appendix D does not include negative values for z, but because normal curves
are symmetrical with respect to a vertical line through the mean, we simply use
the absolute value (positive value) of z for the table. So the area corresponding to
z = - 1.20 is the same as the area corresponding to z = 1.20, which is .3849. And
since the area under the whole normal curve is 1, we conclude that the probability of
a lightbulb chosen at random lasting between 380 and 500 hours is .3849.
Matched Problem 2 Refer to Example 1. What is the probability that a light-

bulb chosen at random lasts between 400 and 500 hours?
The first graphing calculator command in Figure 8A simulates the life expectan-
cies of 100 lightbulbs by generating 100 random numbers from the normal distribu-
tion with m = 500, s = 100 of Example 1. The numbers are stored in list L 1. Note
from Figure 8A that the mean and standard deviation of L 1 are close to the mean and
standard deviation of the normal distribution. From Figure 8B, which shows a histo-
gram of L 1, we note that the empirical probability that a lightbulb lasts between 380
and 500 hours is
11 + 18 + 13
= .42
100
which is close to the theoretical probability of .3849 computed in Example 2.
Figure 8 (A) (B)

Several important properties of a continuous random variable with normal distri-

bution are listed below:
Properties of a Normal Probability Distribution

1. P1a … x … b2 = area under the normal curve from a to b
2. P1 - ∞ 6 x 6 ∞2 = 1 = total area under the normal curve
3. P1x = c2 = 0
In Example 2, what is the probability of a lightbulb chosen at random hav-
ing a life of exactly 621 hours? The area above 621 and below the normal curve at
x = 621 is 0 (a line has no width). Thus, the probability of a lightbulb chosen at ran-
dom having a life of exactly 621 hours is 0. However, if the number 621 is the result
of rounding a number between 620.5 and 621.5 (which is most likely the case), then
the answer to the question is
P1620.5 … x … 621.52 = area under the normal curve from 620.5 to 621.5
The area is found using the procedures outlined in Example 2.
We have just pointed out an important distinction between a continuous random
variable and a discrete random variable: For a probability distribution of a continu-
ous random variable, the probability of x assuming a single value is always 0. On
the other hand, for a probability distribution of a discrete random variable, the prob-
ability of x assuming a particular value from the set of permissible values is usually a
positive number between 0 and 1.
Approximating a Binomial Distribution with a Normal Distribution

You no doubt discovered from problems in Exercises 9.4 that when a binomial ran-
dom variable assumes a large number of values (that is, when n is large), the use of
the probability distribution formula
P1x successes in n trials2 = nCx pxqn - x
becomes tedious. It would be helpful if there was an easily computed approximation
of this distribution for large n. Such a distribution is found in the form of an appropri-
ately selected normal distribution.
To clarify ideas and relationships, let us consider an example of a normal distri-
bution approximation of a binomial distribution with a relatively small value of n.
Then we will consider an example with a large value of n.
EXAMPLE 3 Market Research A credit card company claims that their card is used by 40%
of the people buying gasoline in a particular city. A random sample of 20 gasoline
purchasers is made. If the company’s claim is correct, what is the probability that
(A) From 6 to 12 people in the sample use the card?
(B) Fewer than 4 people in the sample use the card?
SOLUTION We begin by drawing a normal curve with the same mean and standard
deviation as the binomial distribution (Fig. 9). A histogram superimposed on this
normal curve can be used to approximate the histogram for the binomial distribu-
tion. The mean and standard deviation of the binomial distribution are
m = np = 12021.42 = 8 n = sample size
s = 1npq = 112021.421.62 ≈ 2.19 p = .4 1from the 40% claim2

f (x)
.180
.160
.140
.120
.100
.080
.060
.040
.020
.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 x
m
Figure 9
(A) To approximate the probability that 6 to 12 people in the sample use the credit
card, we find the area under the normal curve from 5.5 to 12.5. We use 5.5
rather than 6 because the rectangle in the histogram corresponding to 6 extends
from 5.5 to 6.5. Reasoning in the same way, we use 12.5 instead of 12. To use
Appendix D, we split the area into two parts: A1 to the left of the mean and A2
to the right of the mean. The sketch in Figure 10 is helpful. Areas A1 and A2 are
found as follows:
x1 - m 5.5 - 8
z1 = = ≈ - 1.14 A1 = .3729
s 2.19
x2 - m 12.5 - 8
z2 = = ≈ 2.05 A2 = .4798
s 2.19
Total area = A1 + A2 = .8527
The approximate probability that the sample will contain between 6 and 12 users
of the credit card is .85 (assuming that the firm’s claim is correct).
f (x)
.180
.160
.140
.120
.100
.080 A1 A2
.060
.040
.020
.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 x
x1 m x2
Figure 10
(B) To use the normal curve to approximate the probability that the sample contains
fewer than 4 users of the credit card, we must find the area A1 under the normal
curve to the left of 3.5. The sketch in Figure 11 is useful. Since the total area under
either half of the normal curve is .5, we first use Appendix D to find the area A2
under the normal curve from 3.5 to the mean 8, and then subtract A2 from .5:
x - m 3.5 - 8
z = = ≈ - 2.05 A2 = .4798
s 2.19
A1 = .5 - A2 = .5 - .4798 = .0202
The approximate probability that the sample contains fewer than 4 users of the
credit card is approximately .02 (assuming that the company’s claim is correct).
f (x)
.180
.160
.140
.120
.100
.080
.060
A2
.040 A1
.020
.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 x
x m
Figure 11
Matched Problem 3 In Example 3, use the normal curve to approximate the

probability that in the sample there are
(A) From 5 to 9 users of the credit card.
(B) More than 10 users of the card.
You no doubt are wondering how large n should be before a normal distribution
provides an adequate approximation for a binomial distribution. Without getting too
involved, the following rule of thumb provides a good test:
Use a normal distribution to approximate a binomial distribution only if
the interval 3 m - 3s, m + 3s4 lies entirely in the interval from 0 to n.
Note that in Example 3, the interval 3m - 3s, m + 3s4 = 31.43, 14.574 lies
entirely within the interval from 0 to 20; so the use of the normal distribution is
justified.

(A) Show that if n Ú 30 and .25 … p … .75 for a binomial distribution, then it
passes the rule-of-thumb test.
(B) Give an example of a binomial distribution that passes the rule-of-thumb test but
does not satisfy the conditions of part (A).
EXAMPLE 4 Quality Control A company manufactures 50,000 ballpoint pens each day. The
manufacturing process produces 50 defective pens per 1,000, on average. A random
sample of 400 pens is selected from each day’s production and tested. What is the
probability that the sample contains
(A) At least 14 and no more than 25 defective pens?
(B) 33 or more defective pens?
SOLUTION Is it appropriate to use a normal distribution to approximate this bino-
mial distribution? The answer is yes, since the rule-of-thumb test passes with ease:
50
m = np = 4001.052 = 20 p = = .05
1,000
s = 1npq = 14001.0521.952 ≈ 4.36
3m - 3s, m + 3s4 = 36.92, 33.084
This interval is well within the interval from 0 to 400.
(A) To find the approximate probability of the number of defective pens in a sample
being at least 14 and not more than 25, we find the area under the normal curve
from 13.5 to 25.5. To use Appendix D, we split the area into an area to the left of
the mean and an area to the right of the mean, as shown in Figure 12.
f (x)
A1 A2
0 5 10 15 20 30 35 x
x1 m x2
Figure 12
x1 - m 13.5 - 20
z1 = = ≈ - 1.49 A1 = .4319
s 4.36
x2 - m 25.5 - 20
z2 = = ≈ 1.26 A2 = .3962
s 4.36
Total area = A1 + A2 = .8281
The approximate probability of the number of defective pens in the sample be-
ing at least 14 and not more than 25 is .83.
(B) Since the total area under a normal curve from the mean on is .5, we find the
area A1 (see Fig. 13) from Appendix D and subtract it from .5 to obtain A2.
f (x)
A1
A2
0 5 10 15 20 25 30 35 x
m x
Figure 13
x - m 32.5 - 20
z = = ≈ 2.87 A1 = .4979
s 4.36
A2 = .5 - A1 = .5 - .4979 = .0021 ≈ .002
The approximate probability of finding 33 or more defective pens in the sample
is .002. If a random sample of 400 included more than 33 defective pens, then the
management could conclude that either a rare event has happened and the manu-
facturing process is still producing only 50 defective pens per 1,000, on average, or
that something is wrong with the manufacturing process and it is producing more
than 50 defective pens per 1,000, on average. The company might have a policy of
checking the manufacturing process whenever 33 or more defective pens are found
in a sample rather than believing a rare event has happened and that the manufactur-
ing process is still running smoothly.
Matched Problem 4 Suppose in Example 4 that the manufacturing process

produces 40 defective pens per 1,000, on average. What is the approximate prob-
ability that in the sample of 400 pens there are
(A) At least 10 and no more than 20 defective pens?
(B) 27 or more defective pens?
When to Use the .5 Adjustment

If we are assuming a normal probability distribution for a continuous random
variable (such as that associated with heights or weights of people), then we find
P1a … x … b2, where a and b are real numbers, by finding the area under the cor-
responding normal curve from a to b (see Example 2). However, if we use a normal
probability distribution to approximate a binomial probability distribution, then we
find P1a … x … b2, where a and b are nonnegative integers, by finding the area un-
der the corresponding normal curve from a - .5 to b + .5 (see Examples 3 and 4).
Exercises 9.5
A In Problems 1–6, use Appendix D to find the area under the stan- B In Problems 27–34, consider the normal distribution with mean
dard normal curve from 0 to the indicated measurement. 60 and standard deviation 12. Find the area under the normal
curve and above the given interval on the horizontal axis.
1. 2.00 2. 3.30
27. [48, 60] 28. [60, 84]
3. 1.24 4. 1.08
29. [57, 63] 30. [54, 66]
5. - 2.75 6. - 0.92
31. 1 - ∞ , 54] 32. [63, ∞ 2
In Problems 7–14, use Appendix D to find the area under the standard
normal curve and above the given interval on the horizontal axis. 33. [51, ∞ 2 34. 1 - ∞ , 78]
7. 3- 1, 14 8. 3- 2, 24 In Problems 35–40, discuss the validity of each statement. If the
9. 3- 0.4, 0.74 10. 3- 0.5, 0.34 statement is always true, explain why. If not, give a counterex-
ample.
11. [0.2, 1.8] 12. 3- 2.1, - 0.94
35. The mean of a normal distribution is a positive real number.
13. 1-∞ , 0.14 14. 30.6, ∞ 2
36. The standard deviation of a normal distribution is a positive
In Problems 15–20, given a normal distribution with mean - 15 real number.
and standard deviation 10, find the number of standard deviations
37. If two normal distributions have the same mean and standard
each measurement is from the mean. Express the answer as a
deviation, then they have the same shape.
positive number.
38. If two normal distributions have the same mean, then they
15. - 22 16. 6.4
have the same standard deviation.
17. - 1.8 18. - 13.5
39. The area under a normal distribution and above the horizontal
19. 10.9 20. - 48.6 axis is equal to 1.
In Problems 21–26, given a normal distribution with mean 25 and 40. In a normal distribution, the probability is 0 that a score lies
standard deviation 5, find the area under the normal curve from more than 3 standard deviations away from the mean.
the mean to the indicated measurement.
In Problems 41–48, use the rule-of-thumb test to check whether a
21. 27.2 22. 36.1 normal distribution (with the same mean and standard deviation
as the binomial distribution) is a suitable approximation for the
23. 12.8 24. 18.7
binomial distribution with
25. 28.3 26. 23.9
41. n = 15, p = .7 42. n = 12, p = .6 (B) Use a graphing calculator to generate 120 scores from
the normal distribution with mean 75 and standard
43. n = 15, p = .4 44. n = 20, p = .6
deviation 8. Determine the number of scores x such that
45. n = 100, p = .05 46. n = 200, p = .03 67 … x … 83, and compare your results with the answer
47. n = 500, p = .05 48. n = 400, p = .08 to part (A).
64. (A) If 250 scores are chosen from a normal distribution with
49. The probability of success in a Bernoulli trial is p = .1. mean 100 and standard deviation 10, how many scores x
Explain how to determine the number of repeated trials would be expected to be greater than 110?
necessary to obtain a binomial distribution that passes the
rule-of-thumb test for using a normal distribution as a suit- (B) Use a graphing calculator to generate 250 scores from
able approximation. the normal distribution with mean 100 and standard
deviation 10. Determine the number of scores greater
50. For a binomial distribution with n = 100, explain how to than 110, and compare your results with the answer to
determine the smallest and largest values of p that pass the part (A).
rule-of-thumb test for using a normal distribution as a suit-
able approximation.
C A binomial experiment consists of 500 trials. The probability of
Applications
success for each trial is .4. What is the probability of obtaining the
number of successes indicated in Problems 51–58? Approximate
these probabilities to two decimal places using a normal curve.
(This binomial experiment easily passes the rule-of-thumb test, 65. Sales. Salespeople for a solar technology company have
as you can check. When computing the probabilities, adjust the average annual sales of $200,000, with a standard deviation
intervals as in Examples 3 and 4.) of $20,000. What percentage of the salespeople would be
expected to make annual sales of $240,000 or more? Assume
51. 185–220 52. 190–205 a normal distribution.
53. 210–220 54. 175–185 66. Guarantees. The average lifetime for a car battery is 170
55. 225 or more 56. 212 or more weeks, with a standard deviation of 10 weeks. If the company
guarantees the battery for 3 years, what percentage of the bat-
57. 175 or less 58. 188 or less teries sold would be expected to be returned before the end of
the warranty period? Assume a normal distribution.
To graph Problems 59–62, use a graphing calculator and refer
to the normal probability distribution function with mean m and 67. Quality control. A manufacturing process produces a criti-
standard deviation s: cal part of average length 100 millimeters, with a standard
1 2 2
deviation of 2 millimeters. All parts deviating by more than
f1x2 = e-1x - m2 >2s (1) 5 millimeters from the mean must be rejected. What percent-
s12p
age of the parts must be rejected, on the average? Assume a
59. Graph equation (1) with s = 5 and normal distribution.
(A) m = 10 (B) m = 15 (C) m = 20 68. Quality control. An automated manufacturing process
Graph all three in the same viewing window with produces a component with an average width of 7.55 cen-
timeters, with a standard deviation of 0.02 centimeter. All
Xmin = - 10, Xmax = 40, Ymin = 0, and Ymax = 0.1.
components deviating by more than 0.05 centimeter from the
60. Graph equation (1) with s = 5 and mean must be rejected. What percentage of the parts must be
(A) m = 8 (B) m = 12 (C) m = 16 rejected, on the average? Assume a normal distribution.
Graph all three in the same viewing window with 69. Marketing claims. A company claims that 60% of the
Xmin = - 10, Xmax = 30, Ymin = 0, and Ymax = 0.1. households in a given community use its product. A competi-
tor surveys the community, using a random sample of 40
61. Graph equation (1) with m = 20 and households, and finds only 15 households out of the 40 in
(A) s = 2 (B) s = 4 the sample use the product. If the company’s claim is correct,
Graph both in the same viewing window with what is the probability of 15 or fewer households using the
product in a sample of 40? Conclusion? Approximate a bino-
Xmin = 0, Xmax = 40, Ymin = 0, and Ymax = 0.2. mial distribution with a normal distribution.
62. Graph equation (1) with m = 18 and 70. Labor relations. A union representative claims 60% of the
(A) s = 3 (B) s = 6 union membership will vote in favor of a particular settle-
Graph both in the same viewing window with ment. A random sample of 100 members is polled, and out
of these, 47 favor the settlement. What is the approximate
Xmin = 0, Xmax = 40, Ymin = 0, and Ymax = 0.2.
probability of 47 or fewer in a sample of 100 favoring the
63. (A) If 120 scores are chosen from a normal distribution with settlement when 60% of all the membership favor the settle-
mean 75 and standard deviation 8, how many scores x ment? Conclusion? Approximate a binomial distribution with
would be expected to satisfy 67 … x … 83? a normal distribution.
71. Medicine. The average healing time of a certain type of inci- Harkins. If Harkins’s claim is correct, what is the probability
sion is 240 hours, with standard deviation of 20 hours. What that only 470 or fewer will favor her in a random sample of
percentage of the people having this incision would heal in 1,000? Conclusion? Approximate a binomial distribution
8 days or less? Assume a normal distribution. with a normal distribution.
72. Agriculture. The average height of a hay crop is 38 inches, 77. Grading on a curve. An instructor grades on a curve by
with a standard deviation of 1.5 inches. What percentage of the assuming that grades on a test are normally distributed. If the
crop will be 40 inches or more? Assume a normal distribution. average grade is 70 and the standard deviation is 8, find the test
scores for each grade interval if the instructor assigns grades as
73. Genetics. In a family with 2 children, the probability that follows: 10% A’s, 20% B’s, 40% C’s, 20% D’s, and 10% F’s.
both children are girls is approximately .25. In a random
sample of 1,000 families with 2 children, what is the approxi- 78. Psychology. A test devised to measure aggressive–passive
mate probability that 220 or fewer will have 2 girls? Approxi- personalities was standardized on a large group of people.
mate a binomial distribution with a normal distribution. The scores were normally distributed with a mean of 50 and
a standard deviation of 10. If we designate the highest 10%
74. Genetics. In Problem 73, what is the approximate probabil-
as aggressive, the next 20% as moderately aggressive, the
ity of the number of families with 2 girls in the sample being
middle 40% as average, the next 20% as moderately passive,
at least 225 and not more than 275? Approximate a binomial
and the lowest 10% as passive, what ranges of scores will be
distribution with a normal distribution.
covered by these five designations?
75. Testing. Scholastic Aptitude Tests (SATs) are scaled so that
the mean score is 500 and the standard deviation is 100. What Answers to Matched Problems
percentage of students taking this test should score 700 or
more? Assume a normal distribution. 1. 49.38%
2. .3413
76. Politics. Candidate Harkins claims that she will receive 52% 3. (A) .70 (B) .13
of the vote for governor. Her opponent, Mankey, finds that 4. (A) .83 (B) .004
470 out of a random sample of 1,000 registered voters favor

9.1 Graphing data EXAMPLES
• Bar graphs, broken-line graphs, and pie graphs are used to present visual interpretations or com- Ex. 1, p. 483
parisons of data. Large sets of quantitative data can be organized in a frequency table, generally
constructed by choosing 5 to 20 class intervals of equal length to cover the data range. The number
of measurements that fall in a given class interval is called the class frequency, and the set of all such
frequencies associated with their respective classes is called a frequency distribution. The relative
frequency of a class is its frequency divided by the total number of items in the data set.
• A histogram is a vertical bar graph used to represent a frequency distribution. A frequency polygon Ex. 2, p. 484
is a broken-line graph obtained by joining successive midpoints of the tops of the bars in a histogram.
A cumulative frequency polygon, or ogive, is obtained by plotting the cumulative frequency over the
upper boundary of the corresponding class.
9.2 Measures of Central Tendency
• The mean of a set of quantitative data is the sum of all the measurements in the set divided by the total Ex. 1, p. 491
number of measurements in the set.
• The mean for data grouped into classes is a weighted average of the midpoints of the class intervals. Ex. 2, p. 492
• When a data set has n measurements and these measurements are arranged in ascending or descending Ex. 3, p. 494
order, the median is the middle measurement when n is odd and the mean of the two middle measure-
ments when n is even.
• The median for grouped data with no classes of frequency 0 is the number such that the histogram Ex. 4, p. 495
has the same area to the left of the median as to the right of the median.
• The mode is the most frequently occurring measurement in a data set. Ex. 5, p. 496
9.3 Measures of Dispersion

• The variance and the standard deviation of a set of ungrouped measurements indicate how the data is Ex. 1, p. 502
dispersed relative to the mean. The same can be said about a set of grouped data. Ex. 2, p. 503
9.4 Bernoulli Trials and Binomial Distributions
• A sequence of experiments is called a sequence of Bernoulli trials, or a binomial experiment, if
1. Only two outcomes are possible in each trial. Ex. 1, p. 507
2. The probability of success p is the same for each trial. Ex. 2, p. 508
3. All trials are independent. Ex. 3, p. 509
• If the random variable Xn represents the number of successes in n Bernoulli trials, then the probability Ex. 5, p. 511
distribution of Xn is the binomial distribution given by P1Xn = x2 = nCx pxq n - x, x = 0, 1, c, n.
• The mean and standard deviation of a binomial distribution are given by the formulas m = np and Ex. 6, p. 512
s = 1npq, respectively.
• Normal curves are bell-shaped continuous curves that approximate the relative frequency distributions
of many different types of measurements.
• The probability that a normally distributed measurement lies between a and b, denoted P1a … x … b2, Ex. 1, p. 518
is equal to the area under the normal curve from a to b. Ex. 2, p. 519
• To approximate a binomial distribution that is associated with a sequence of n Bernoulli trials, each having Ex. 3, p. 520
probability of success p, use a normal distribution with m = np and s = 1npq. Ex. 4, p. 522
Review Exercises
Work through all the problems in this chapter review and check 3. (A) Draw a histogram for the binomial distribution
your answers in the back of the book. Answers to all review prob- P1x2 = 3Cx 1.42 x 1.62 3 - x
(B) What are the mean and standard deviation?
up, review appropriate sections in the text. 4. For the set of sample measurements 1, 1, 2, 2, 2, 3, 3, 4, 4, 5,
find the
A 1. Use a bar graph and a broken-line graph to graph the data on
voter turnout, as a percentage of the population eligible to vote, (A) Mean (B) Median
in U.S. presidential elections. (Source: U.S. Elections Project) (C) Mode (D) Standard deviation
Voter Turnout in U.S. Presidential Elections 5. If a normal distribution has a mean of 100 and a standard
Percentage of deviation of 10, then
Year Eligible Voters (A) How many standard deviations is 118 from the mean?
1996 53 (B) What is the area under the normal curve between the
2000 55 mean and 118?
2004 61 B 6. Given the sample of 25 quiz scores listed in the following
2008 62 table from a class of 500 students:
2012 59 (A) Construct a frequency table using a class interval of
2016 60 width 2, starting at 9.5.
2. Use a pie graph to graph the data on educational attainment in the (B) Construct a histogram.
U.S. population of adults 25 years of age or older. (Source: U.S.
(C) Construct a frequency polygon.
Census Bureau, 2015 American Community Survey)
(D) Construct a cumulative frequency and relative cumula-
Educational Attainment in the United States tive frequency table.
Percentage (E) Construct a cumulative frequency polygon.
Attainment (25 Years or Older) Quiz Scores
Less than high school diploma 13 14 13 16 15 17
High school diploma 28 19 15 14 17 15
Some college or associate’s degree 29 15 13 12 14 14
Bachelor’s degree 19 12 14 13 11 15
Graduate or professional degree 12 16 14 16 17 14
7. For the set of grouped sample data given in the table, 17. Two dice are rolled three times. What is the probability of
(A) Find the mean. getting a sum of 7 at least once?
(B) Find the standard deviation. 18. Ten students take an exam worth 100 points.
(C) Find the median. (A) Construct a hypothetical set of exam scores for the ten
students in which both the median and the mode are
Interval Frequency 30 points higher than the mean.
0.5–3.5 1 (B) Could the median and mode both be 50 points higher
3.5–6.5 5 than the mean? Explain.
6.5–9.5 7
19. In the last presidential election, 39% of a city’s registered
9.5–12.5 2 voters actually cast ballots.
8. (A) Construct a histogram for the binomial distribution
(A) In a random sample of 20 registered voters from that
P1x2 = 6Cx 1.52 x 1.52 6 - x city, what is the probability that exactly 8 voted in the
(B) What are the mean and standard deviation? last presidential election?
9. What are the mean and standard deviation for a binomial (B) Verify by the rule-of-thumb test that the normal distribu-
distribution with p = .6 and n = 1,000? tion with mean 7.8 and standard deviation 2.18 is a good
approximation of the binomial distribution with n = 20
In Problems 10 and 11, discuss the validity of each statement. If the
and p = .39.
statement is always true, explain why. If not, give a counterexample.
(C) For the normal distribution of part (B), P1x = 82 = 0.
10. (A) If the data set x1, x2, c, xn has mean x, then the data Explain the discrepancy between this result and your
set x1 + 5, x2 + 5, c, xn + 5 has mean x + 5. answer from part (A).
(B) If the data set x1, x2, c, xn has standard deviation s, 20. A random variable represents the number of wins in a 12-
then the data set x1 + 5, x2 + 5, c, xn + 5 has stan- game season for a football team that has a probability of .9 of
dard deviation s + 5. winning any of its games.
11. (A) If X represents a binomial random variable with mean m, (A) Find the mean and standard deviation of the random
then P1X Ú m2 = .5. variable.
(B) If X represents a normal random variable with mean m, (B) Find the probability that the team wins each of its 12
then P1X Ú m2 = .5. games.
(C) The area of a histogram of a binomial distribution is equal (C) Use a graphing calculator to simulate 100 repetitions
to the area above the x axis and below a normal curve. of the binomial experiment associated with the random
12. If the probability of success in a single trial of a binomial variable, and compare the empirical probability of a
experiment with 1,000 trials is .6, what is the probability of perfect season with the answer to part (B).
obtaining at least 550 and no more than 650 successes in
1,000 trials? (Hint: Approximate with a normal distribution.)
13. Given a normal distribution with mean 50 and standard de-
viation 6, find the area under the normal curve: Applications
(A) Between 41 and 62
21. Retail sales. The daily number of bad checks received by a
(B) From 59 on large department store in a random sample of 10 days out of
14. A data set is formed by recording the sums of 100 rolls of a the past year was 15, 12, 17, 5, 5, 8, 13, 5, 16, and 4. Find the
pair of dice. A second data set is formed by again rolling a (A) Mean
pair of dice 100 times but recording the product, not the sum,
of the two numbers. (B) Median
(A) Which of the two data sets would you expect to have the (C) Mode
smaller standard deviation? Explain.
(D) Standard deviation
(B) To obtain evidence for your answer to part (A), use a graph-
ing calculator to simulate both experiments, and compute 22. Preference survey. Find the mean, median, and/or mode,
the standard deviations of each of the two data sets. whichever are applicable, for the following employee cafete-
ria service survey:
C 15. For the sample quiz scores in Problem 6, find the mean and
standard deviation using the data
Drink Ordered with Meal Number
(A) Without grouping.
Coffee 435
(B) Grouped, with class interval of width 2, starting at 9.5.
Tea 137
16. A fair die is rolled five times. What is the probability of Milk 298
rolling
Soft drink 522
(A) Exactly three 6’s? Milk shake 392
(B) At least three 6’s?
23. Plant safety. The weekly record of reported accidents in a 25. Market research. A newspaper publisher claims that 70%
large auto assembly plant in a random sample of 35 weeks of the people in a community read their newspaper. Doubting
from the past 10 years is listed below: the assertion, a competitor randomly surveys 200 people in
the community. Based on the publisher’s claim (and assum-
34 33 36 35 37 31 37 ing a binomial distribution),
39 34 35 37 35 32 35
(A) Compute the mean and standard deviation.
33 35 32 34 32 32 39
34 31 35 33 31 38 34 (B) Determine whether the rule-of-thumb test warrants the
36 34 37 34 36 39 34 use of a normal distribution to approximate this binomial
distribution.
(A) Construct a frequency and relative frequency table using (C) Calculate the approximate probability of finding at least
class intervals of width 2, starting at 29.5. 130 and no more than 155 readers in the sample.
(B) Construct a histogram and frequency polygon. (D) Determine the approximate probability of finding 125 or
(C) Find the mean and standard deviation for the grouped fewer readers in the sample.
data. (E) Use a graphing calculator to graph the relevant normal
24. Personnel screening. The scores on a screening test for distribution.
new technicians are normally distributed with mean 100 and 26. Health care. A small town has three doctors on call for
standard deviation 10. Find the approximate percentage of emergency service. The probability that any one doctor will
applicants taking the test who score be available when called is .90. What is the probability that at
(A) Between 92 and 108 least one doctor will be available for an emergency call?
(B) 115 or higher

10 Games and Decisions
10.1 Strictly Determined
Games
Introduction
Game theory is a relatively new branch of mathematics designed to help
10.2 Mixed-Strategy Games
people in conflict situations determine the best course of action out of several
10.3 Linear Programming possible choices. The theory is applicable to some games, but more impor-
and 2 * 2 Games: tantly, it has been applied with moderate success to decision-making problems
in economics, business, psychology, sociology, warfare, and political science.
10.4 Linear Programming For example, should a farmer plant corn or soybeans? See Problems 51 and
and m * n Games: 52 in the Chapter 10 Review Exercises.
Simplex Method and The theory had its beginnings in the 1920s, but its greatest advance oc-
the Dual Problem
curred in 1944, when John von Neumann and Oskar Morgenstern published
their landmark book, Games and Economic Behavior. Fifty years later, in 1994,
a Nobel Prize in Economic Sciences was awarded to John Nash, John Harsanyi,
and Reinhard Selten for their work in game theory.
Game theory provides an excellent review of many of the topics studied in
preceding chapters. Linear systems, matrices, probability, expected value, and
linear programming are all used in the development of this subject.
G1
G2 CHAPTER 10 Games and Decisions
10.1 Strictly Determined Games

■■ Strictly Determined Matrix Games The best way to start this discussion is with an example. Out of this example will
■■ Nonstrictly Determined Matrix
evolve basic definitions, theorems, and methodology.
Games Consider two stores that sell HDTVs, store R and store C. Each is trying to de-
cide how to price a particular model. A market research firm supplies the following
information:
Store C
$499 $549
$499 55% 70% (1)
Store R J R
$549 40% 66%
The matrix entries indicate the percentage of the business that store R will receive.
That is, if both stores price their HDTV at $499, store R will receive 55% of all the
business (store C will lose 55% of the business but will get 45%). If store R chooses
a price of $499 and store C chooses $549, store R will receive 70% of the business
(store C will lose 70% of the business but will get 30%), and so on. Each store can
choose its own price but cannot control the price of the other. The object is for each
store to determine a price that will ensure the maximum possible business in this
competitive situation.
This marketing competition may be viewed as a game between store R and store
C. A single play of the game requires store R to choose (play) row 1 or row 2 in ma-
trix (1) (that is, price its HDTV at either $499 or $549) and simultaneously requires
store C to choose (play) column 1 or column 2 (that is, price its HDTV at either $499
or $549). It is common to designate the person(s) choosing the rows by R, for row
player, and the person(s) choosing the columns by C, for column player. Each entry
in matrix (1) is called the payoff value for a particular pair of moves by R and C.
Matrix (1) is called a game matrix or a payoff matrix. This game is a two-person
zero-sum game because there are only two players and one player’s win is the other
player’s loss.
Strictly Determined Matrix Games

Actually, any m : n matrix may be considered a two-person zero-sum matrix
game in which player R chooses (plays) any one of m rows and player C simultane-
ously chooses (plays) any one of n columns. For example, the 3 * 4 matrix
0 6 -2 -4
C 5 2 1 3S (2)
-8 -1 0 20
may be viewed as a matrix game where R has three moves and C has four moves. If
R plays row 2 and C plays column 4, then R wins 3 units. If, however, R plays row 3
and C plays column 1, then R “wins” - 8 units; that is, C wins 8 units.
Negative entries in the payoff matrix indicate a win for C, and positive
entries indicate a win for R.
How should R and C play in matrix game (2)? If R is a little greedy, row 3 might
be chosen because the largest payoff is 20 units. But we assume that C would likely
avoid column 4 (not wanting to lose 20 units). If C plays column 1, then 8 units might
be won. But then R might play row 2, anticipating C’s thinking, and C would not win
after all. Is there a best play for each? To help us unravel this problem, we state a
fundamental principle of game theory:
SECTION 10.1 Strictly Determined Games G3
FUNDAMENTAL PRINCIPLE OF GAME THEORY

1. A matrix game is played repeatedly.
2. Player R tries to maximize winnings.
3. Player C tries to minimize losses.
Player R, being conservative, thinks in terms of the worst that could happen
for each row choice and chooses a row that has the largest minimum payoff. This
provides a security level that is guaranteed irrespective of C’s choices. Returning to
matrix game (2), we find that the worst that could happen in row 1 is a 4-unit loss, in
row 2 a 1-unit gain, and in row 3 an 8-unit loss. Each of these values is circled below.
0 6 -2 -4
C 5 2 1 3S
-8 -1 0 20
The best approach (strategy) for R is to select the row with the largest of these mini-
mum values—that is, row 2. With this choice, a win of at least 1 unit is guaranteed for
R no matter what C does!
Similarly, C puts a square around the maximum value in each column to identify
the worst situation that could happen for each column choice. (Remember that a posi-
tive number indicates a loss for C.)
0 6 -2 -4 Circles mark the minimum value in each row,
C 5 2 1 3S and squares mark the maximum value in
each column.
-8 -1 0 20
We see that C’s best approach (strategy) is to select the column with the smallest
of these maximum values—that is, column 3. By choosing column 3, C establishes a
security level of a loss of 1 unit irrespective of R’s choices, and this is the best that C
can do if R continues to make the best moves.
The entry 1 in the second row and third column (enclosed by a circle and a
square) is both the largest of the row minimums and the smallest of the column maxi-
mums. This means that R should always play the second row and C should always
play the third column. The result is a win of 1 unit for R every time.
These are the best (optimal) strategies for both R and C. If C keeps playing the
third column and R decides to change from the second row, then R’s wins cannot in-
crease. Similarly, if R continues to play the second row and C deviates from the third
column, then C’s losses cannot decrease. The game is said to be strictly determined
in that R must always play row 2 to maximize winnings, and C must always play col-
umn 3 to minimize losses.
The payoff value surrounded by both a circle and a square is called a saddle
value. To see why, drop all the elements in the preceding matrix except those in the
second row and third column to obtain
-2
C5 2 1 3S
0
In one direction, 1 is a minimum; in the other direction, 1 is a maximum. The form is
characteristic of a horse saddle.
DEFINITION Saddle Value

A saddle value is a payoff value that is simultaneously a row minimum and a column
maximum of a payoff matrix.
PROCEDURE Locating Saddle Values

Step 1 Circle the minimum value in each row (it may occur in more than one place).
Step 2 Place squares around the maximum value in each column (it may occur in
more than one place).
Step 3 Any entry with both a circle and a square around it is a saddle value.

Suppose that a and k are both saddle values of the matrix
a b c d
A = £e f g h§
i j k l
(A) Show that a must equal k after explaining why each of the following must be
true: a … c, c … k, a Ú i, and i Ú k.
(B) Show that c and i must equal k as well.
Explore and Discuss 1 indicates that a payoff matrix cannot have two different
numbers as saddle values. We state this result as Theorem 1. The same number, how-
ever, might well appear several times as a saddle value in a payoff matrix.
THEOREM 1 Equality of Saddle Values

If a payoff matrix has saddle values x and y, then x = y.
Theorem 1 implies that a payoff matrix either has exactly one saddle value (pos-
sibly appearing several times as a saddle value) or no saddle value. A matrix game
with a saddle value is called strictly determined. A matrix game with no saddle value
is called nonstrictly determined. The 3 * 3 matrix in which each entry is 1 is an ex-
ample of a strictly determined matrix game (1 appears nine times as a saddle value).
The 3 * 3 identity matrix is an example of a nonstrictly determined matrix game.
DEFINITION Strictly Determined Matrix Games

A matrix game is said to be strictly determined if it has a saddle value. In a
strictly determined game, optimal strategies are as follows:
• R should choose any row containing a saddle value.
• C should choose any column containing a saddle value.
A saddle value is called the value of a strictly determined game. The game is fair
if its value is zero.
Note: In a strictly determined game (assuming that both players play their opti-
mal strategy), knowledge of an opponent’s move provides no advantage since the
payoff will always be a saddle value.
EXAMPLE 1 Finding Saddle Values and Optimal Strategies Two large shopping centers
have competing home-improvement discount stores, Riteway (R) in one center and
Cashmart (C ) in the other. Every week, each store chooses one, and only one, of the
following means of promotion: TV, radio, newspaper, or mail. A marketing research
company provided the following payoff matrix, which indicates the percentage of
market gain or loss for each choice of action by R and C (we assume that any gain
by R is a loss by C, and vice versa):
C
TV Radio Paper Mail
TV 0 -2 -2 2
Radio 1 2 1 3
R D T
Paper - 1 0 0 1
Mail 1 2 1 2
(A) Locate the saddle values.
(B) Find optimal strategies for R and C.
(C) Find the value of the game.
SOLUTION
(A) Follow the three steps listed in the procedure on page G4 for locating saddle values:
C
TV Radio Paper Mail
TV 0 -2 -2 2
Radio 1 2 1 3
R D T
Paper - 1 0 0 1
Mail 1 2 1 2
The entry 1 appears four times as a saddle value.
(B) Optimal strategies for R and C are to choose rows and columns, respectively,
that contain a saddle value.
• Optimal strategy for R: Choose radio or mail each week.
• Optimal strategy for C: Choose TV or newspaper each week.
(C) The value of the game is 1, and store R has the advantage. (When both stores
use optimal strategies, store R will gain 1% of the market at store C’s expense
of losing 1%.)
Matched Problem 1 Repeat Example 1 for the HDTV game matrix discussed
at the beginning of this section:
C
$499 $549
$499 55% 70%
R c d
$549 40% 66%
Nonstrictly Determined Matrix Games

Not all matrix games are strictly determined; that is, many matrix games do not have sad-
dle values. Consider the classic penny-matching game: We have two players, player R and
player C. Each player has a penny, and they simultaneously choose to show the side of
the coin of their choice 1H = heads, T = tails2. If the pennies match, R wins (C loses)
1¢. If the pennies do not match, R loses (C wins) 1¢. In terms of a game matrix, we have
Player C
H T
H 1 -1
Player R c d
T -1 1
Testing this game matrix for saddle values, we obtain
Player C
H T
H 1 -1
Player R c d
T -1 1
No entry is enclosed in both a circle and a square; hence, there is no saddle value and
the game is called nonstrictly determined. A minimum of - 1 occurs in each row;
thus, R could play either. Similarly, a maximum of 1 occurs in each column, and C
could play either. In a nonstrictly determined game, knowledge of the other player’s
move would certainly be very useful. For example, if C knew that R was going to
play row 1 (heads), then C would obviously play column 2 (tails) to win 1¢.
In this game, is there an optimal strategy for each player? In general, for non-
strictly determined two-person zero-sum games, are there optimal strategies for each
player? Surprisingly, the answer turns out to be yes—this is the subject of Sections
10.2 through 10.4.
EXAMPLE 2 Distinguishing between Strictly and Nonstrictly Determined Games Determine

whether the following matrix games are strictly determined:
2 -1 -8
-1 2 -3 4
A = c d B = C 4 8 10 S
5 -2 3 0
-5 -3 0
SOLUTION Using the circle and square technique discussed previously, we obtain
2 -1 -8
-1 2 -3 4
A = c d
B = C 4 8 10 S
5 -2 3 0
-5 -3 0
Matrix game A has no saddle value and thus is nonstrictly determined. Matrix game
B has a saddle value (4 in row 2 and column 1) and hence is strictly determined.
Matched Problem 2 Determine which of the matrix games below are non-
strictly determined:
-1 3
2 -1 4
0 2
A = D T B = C -3 0 5S
-2 -4
-6 10 -8
-1 0
Exercises 10.1
A In Problems 1–8, is the matrix game strictly determined? -5 3 1 -2 4 -1
11. £ 5 4 8§ 12. £ 0 2 3§
-2 -1 5 -3
1. c d 2. c d -2 -1 -3 -1 0 4
3 1 -1 2
7 2 -3 -2 3 -2 0 2 0 -3 0 2
3. c d 4. c d
-5 0 1 -2 5 -2 13. £ - 3 -2 -1 § 14. £ 0 1 0§
0 -1 4 -1 1 -3 -1 -5 0 -2 2 -1
5. C 3 -2 -1 S 6. C 2 7 1S -2 7 3 -4 -2 1
1 -1 2 0 4 -1 15. £ - 1 -3 0§ 16. £ - 1 0 -1 §
4 -2 5 -1 5 3 5 4 3 4 2 1
7. £ - 3 1 0§ 8. £ 4 - 2 - 4 §
In Problems 17–32, for each matrix game that is strictly deter-
-1 2 3 6 8 7
mined (if it is not strictly determined, say so)
In Problems 9–16, the matrix for a strictly determined game is
given. Find the value of the game. Is the game fair? (A) Locate the saddle values.
(B) Find optimal strategies for R and C.
1 0 1 2
9. c d 10. c d
-2 -1 -4 3 (C) Find the value of the game.
4 -3 -2 3
17. c
5 4
d 18. c
-3 6
d Applications
43. Price war. A small town on a major highway has only two
-1 0 2 0
19. c d 20. c d gas stations: station R, a major brand station, and station C,
3 -1 0 -4 an independent. A market research firm provided the follow-
ing payoff matrix, where each entry indicates the percentage
-1 -1 3 -2 of customers who go to station R for the indicated prices per
21. c d 22. c d
-1 -1 -5 1 gallon of unleaded gasoline. Find saddle values and optimal
strategies for each company.
2 2 4 4 Station C
23. c d 24. c d
2 5 -2 4 $1.35 $1.40
$1.40 50% 70%
2 -1 -5 Station R c d
1 0 3 1 $1.45 40% 50%
B 25. C 1 0 3S 26. c d
-5 -2 4 -3
-3 -7 8 44. Investment. Suppose that you want to invest $10,000 for a
period of 5 years. After getting financial advice, you come
3 -2
1 -3 5 up with the following game matrix, where you (R) are play-
1 5 ing against the economy (C). Each entry in the matrix is the
27. D T 28. C - 2 1 6S
-4 0 expected payoff (in dollars) after 5 years for an investment of
3 -4 0
5 -3 $10,000 in the corresponding row designation, with the future
state of the economy in the corresponding column designa-
3 -1 4 -7 tion. (The economy is regarded as a rational player who can
1 -2 0 3
1 0 2 3 make decisions against the investor—in any case, the investor
29. D T 30. C - 5 0 -1 8S
5 -2 -3 0 would like to do the best possible, irrespective of what hap-
4 1 1 2 pens to the economy.) Find saddle values and optimal strate-
3 0 1 5
gies for each player.
0 4 -8 -3 -1 9 -1 -1
2 5 3 2 -2 4 -3 -2 Economy C
31. D T 32. D T Fall No change Rise
1 -3 -2 -9 -1 5 -1 -1
2 4 7 2 -3 0 -2 -4 5-year CD 5,870 5,870 5,870
Investor R Blue-chip stock £ - 2,000 4,000 7,000 §
33. For the matrix game of Problem 31, would you rather be
player R or player C? Explain. Speculative stock - 5,000 2,000 10,000
34. For the matrix game of Problem 32, would you rather be 45. Store location. Two competitive pet shops want to open
player R or player C? Explain. stores at Lake Tahoe, where there are currently no pet shops.
The following figure shows the percentages of the total Tahoe
C In Problems 35–40, discuss the validity of each statement. If the population serviced by each of the three main business centers.
statement is always true, explain why. If not, give a counterexample. If both shops open in the same business center, then they split
35. There exists a payoff matrix that has exactly two saddle values. all the business equally; if they open in two different centers,
then they each get all the business in the center in which they
36. There exists a payoff matrix having a saddle value that open plus half the business in the third center. Where should
appears exactly twice. the two pet shops open? Set up a game matrix and solve.
37. The smallest entry in any payoff matrix is a saddle value.
38. The largest entry in any payoff matrix is a saddle value.
Incline
39. If a payoff matrix has a row consisting of all 0’s and a col- Village
umn consisting of all 0’s, then the game is fair. 30%
40. If a strictly determined matrix game is fair, then at least one Tahoe City
of the payoffs is 0. 20%
Lake Tahoe
41. Is there a value of m such that the following is not a strictly
determined matrix game? Explain.
-3 m
c d
0 1
42. If M is a 2 * 2 matrix game and both entries in one row are South
the same, try to find values for the other row so that the game Lake Tahoe
is not strictly determined. What is your conclusion? 50%
46. Store location. Two competing auto parts companies (R and Answers to Matched Problems
C) are trying to decide among three small towns (E, F, and G) 1. (A) The 55% in the first row and first column is the only
for new store locations. All three towns have the same business saddle value.
potential. If both companies operate in the same town, they (B) Optimal strategy for R: Play row 1 ($499 price).
split the business evenly (payoff is 0 to both). If, however, they (C) Optimal strategy for C: Play column 1 ($499 price).
operate in different towns, the store that is closer to the third (D) The value of the game is 55%; so store R has the
town will get all of that town’s business. For example, if R op- advantage. (If the value had been 50%, neither store
erates in E and C in F, the payoff is 1 (R has gained one town). would have an advantage; if it had been 45%, store
If, on the other hand, R operates in F and C in E, the payoff is C would have the advantage.)
- 1 (R has lost one town to C). Write the payoff matrix, find all 2. B is nonstrictly determined.
saddle values, and indicate optimal strategies for both stores.
G
10 kilometers 14 kilometers
E 20 kilometers F

■■ Nonstrictly Determined Games: Two-person zero-sum games can be divided into two classes:
Example
1. Strictly determined games
■■ Pure and Mixed Strategies 2. Nonstrictly determined games
■■ Expected Value of a Game In the preceding section, we found that in a strictly determined game, both players
■■ Fundamental Theorem of always choose rows and columns that contain a saddle value. The game is completely
Game Theory open in the sense that knowledge of an opponent’s strategy will be of no advantage
to either player.
■■ Solution to a 2 * 2 Matrix Game
■■ Recessive Rows and Columns Nonstrictly Determined Games: Example
Now we consider some interesting nonstrictly determined games. As before, we start
with an example and generalize from our observations.
There are many variations of the two-finger Morra game. Basically, the game
involves two players who simultaneously show one or two fingers and agree to a
payoff for each particular combined outcome. Let us consider the following variation
of the game, where the payoff (in dollars) is the sum of the fingers to R if the sum is
even and the sum of the fingers to C if the sum is odd:
Player C
1 finger 2 fingers
Player R 1 finger 2 -3
c d
2 fingers -3 4
Would you rather be a row player or a column player, or does it matter? (Think about
this for a moment before proceeding.)
To answer this question, start by checking to see whether the game is strictly
determined. Using the circle and square method described in the preceding section,
we have
2 -3
c d
-3 4
Since no payoff has both a circle and a square around it, there is no saddle value;
hence, the game is not strictly determined. How should the players play?
If player R continued to play row 2 (because of the $4 payoff), it would not take
C long to detect R’s strategy, and C would obviously play column 1. Then R would
play row 1 and C would have to shift to column 2, and so on. It appears that neither
SECTION 10.2 Mixed-Strategy Games G9
player should continue playing the same row or same column but should play each
row and each column in some mixed pattern unknown to the other player.
How should a mixed pattern be chosen? The best way to choose a mixed pattern
is to use a probability distribution and a chance device that produces this distribution.
For example, R might choose row 1 with probability 14 and row 2 with probability 34.
This can be accomplished by placing 1 white marble and 3 black marbles in an urn and
drawing one at random, letting white represent row 1 and black represent row 2. (The
drawn marble would be replaced after each play.) In this way, neither the player nor the
opponent would know which row was to be played until a marble was drawn. But in
the long run, R would play row 1 one-fourth of the time and row 2 three-fourths of the
time. Similarly, player C might choose column 1 with probability 35 and column 2 with
probability 25 by random drawings from an urn containing 3 white and 2 black marbles.
Pure and Mixed Strategies

At this time it is useful to carefully define the word strategy.
DEFINITION Strategies for R and C

Given the game matrix
a b
M = c d
c d
R’s strategy is a probability row C’s strategy is a probability
matrix: column matrix:
p1 Ú 0 q1 Ú 0
q1
P = 3p1 p2 4 p2 Ú 0 Q = c d q2 Ú 0
q2
p1 + p2 = 1 q1 + q2 = 1
In the previous example, R would be using the strategy

P = 3 14 3
4 4 Row matrix
and C would be using the strategy
3
Q = c 52 d Column matrix
5
The reasons for using row and column matrices for strategies will be made clear shortly.
If one of the elements in P (or Q) is 1 and the other is 0, then the strategy is
a pure strategy. If a strategy is not a pure strategy, then it is a mixed strategy.
So P = 30 14 is a pure strategy, and R would play row 2 for every play. On the
other hand, P = 3 14 34 4 is a mixed strategy, indicating that R plays row 1 with
probability 14 and row 2 with probability 34 .
In the two-finger Morra game described earlier, we found that pure strategies were
not the best choice (pure strategies are always used in strictly determined games, and
the preceding game is not strictly determined). This indicates that mixed strategies
should be used. But what mixed strategies? Are there optimal mixed strategies for
both players? The answer happens to be yes, and now we discuss a way to find them.
Expected Value of a Game

We will use the idea of expected value discussed in Section 7.5. What is the ex-
pected value of the two-finger Morra game for R if strategies
q1 .6
P = 3p1 p2 4 = 3.25 .754 and Q = c d = c d
q2 .4
are used by R and C, respectively? There are four possible outcomes (payoffs) for R,
and each has a probability of occurrence that can be determined. Starting with
a b 2 -3
M = c d = c d
c d -3 4
we see that the probability of payoff a occurring is the probability associated with
R playing row 1 and C playing column 1. Since R’s and C’s plays are independent
(neither knows what the other will play), the probability of a occurring is the product
p1q1. Similarly, the probabilities of payoffs b, c, and d are p1q2, p2q1, and p2q2, re-
spectively. So using the definition of expected value from Section 7.5, we see that the
expected value of the game for R, denoted as E(P, Q), is given by
E1P, Q2 = ap1q1 + bp1q2 + cp2q1 + dp2q2 (1)

= 1221.2521.62 + 1 - 321.2521.42 + 1 - 321.7521.62 + 1421.7521.42
= - 0.15
This means that in the long run, with R and C using the indicated strategies, R will
lose 15¢ per game, on average, and C will win 15¢ per game, on average. Could R do
better by using different probabilities p1 and p2?
Player R wants to find a strategy that gives the largest expected value
irrespective of C’s choice of strategies. On the other hand, C wants to
find a strategy that would give R the smallest expected value irrespec-
tive of R’s choice of strategies.
To find a better strategy for R, we rewrite E(P, Q) as follows, treating p1, p2, q1,
and q2 as variables:
E1P, Q2 = ap1q1 + bp1q2 + cp2q1 + dp2q2 Let a = 2, b = - 3, c = - 3, d = 4.

= 2p1q1 - 3p1q2 - 3p2q1 + 4p2q2 Let p2 = 1 - p1, q 2 = 1 - q1.
= 2p1q1 - 3p1 11 - q1 2 - 311 - p1 2q1 + 411 - p1 211 - q1 2 Distribute.
= 2p1q1 - 3p1 + 3p1q1 - 3q1 + 3p1q1 + 4 - 4p1 - 4q1 + 4p1q1
Combine like terms.
= 12p1q1 - 7p1 - 7q1 + 4 Commute inner terms.
= 12p1q1 - 7q1 - 7p1 + 4 Factor out q1.
= 112p1 - 72q1 - 7p1 + 4
If R chooses p1 so that 12p1 - 7 = 0 (that is, p1 = 7>12), then
E1P, Q2 = 0 # q1 - 7 # 7
+ 4 = -
49
+
48
= -
1
12 12 12 12
regardless of the value of q1. Note that - 1>12 ≈ - 0.0833, so this is an improve-
ment for R, who now loses a little more than 8¢ per game (instead of losing 15¢
per game). Furthermore, this is the best that R can do. For if R were to choose p1 so
that 12p1 - 7 7 0 (that is, p1 7 7>12), then C could choose q1 = 0 to guarantee a
smaller expected value for R:
E1P, Q2 = 112p1 - 72q1 - 7p1 + 4 Let q1 = 0.
= 0 - 7p1 + 4 p1 7 7>12 implies - 7p1 6 - 49>12.

49 1
6 - + 4 = -
12 12
Similarly, if R were to choose p1 so that 12p1 - 7 6 0 (that is, p1 6 7>12), then C

could choose q1 = 1 to guarantee a smaller expected value for R:
E1P, Q2 = 112p1 - 72q1 - 7p1 + 4 Let q1 = 1.
= 12p1 - 7 - 7p1 + 4 Combine like terms.
= 5p1 - 3 p1 6 7/12 implies 5p1 6 35>12.
35 1
6 - 3 = -
12 12
To summarize, the optimal strategy for R in the two-finger Morra game is to choose
p1 = 7>12, so
7 5
P = c d
12 12
By a similar argument, the optimal strategy for C is to choose q1 = 7>12, so
7 5
Q = c d
12 12
(In general, P and Q are different, but they are equal here because of the symmetry of
the payoff matrix M.)
Our results for the two-finger Morra game can be generalized to give the funda-
mental theorem of game theory. But before we state that theorem, here is a conve-
nient way of denoting the expected value E(P, Q)—this will clarify the reason why
we represented P and Q as row and column matrices, respectively:
THEOREM 1 Expected Value of a Matrix Game for R

For the matrix game
a b
M = c d
c d
and strategies
q1
P = 3p1 p2 4 Q = c d
q2
for R and C, respectively, the expected value of the game for R is given by
E1P, Q2 = PMQ
To show that E1P, Q2 = PMQ, calculate the matrix product PMQ to obtain
PMQ = ap1q1 + bp1q2 + cp2q1 + dp2q2
But from equation (1),
E1P, Q2 = ap1q1 + bp1q2 + cp2q1 + dp2q2
So E1P, Q2 = PMQ. (The details of expanding PMQ are left to Problem 41 in
Exercises 10.2.)
CONCEPTUAL INSIGHT
1. After multiplying, PMQ is a 1 * 1 matrix, which is usually written without
brackets.
2. The relation E1P, Q2 = PMQ holds for arbitrary m * n matrix games
(including strictly determined games), where P is a 1 * m row matrix and Q is
an n * 1 column matrix.
Fundamental Theorem of Game Theory

We have shown that both R and C have optimal strategies in the two-finger Morra
game: R should play row 1 with probability 7>12 and row 2 with probability 5>12;
and C should play column 1 with probability 7>12 and column 2 with probability
5>12. The fundamental theorem of game theory, which we state without proof,
guarantees that R and C have optimal strategies in any m * n matrix game.
THEOREM 2 Fundamental Theorem of Game Theory†

For every m * n matrix game M, there exist strategies P* and Q* (not necessarily
unique) for R and C, respectively, and a unique number v such that
P*MQ Ú v (2)
for every strategy Q of C and
PMQ* … v (3)
for every strategy P of R.
The number v is called the value of the game and is the security level for both R
and C. If v = 0, then the game is said to be fair. Furthermore, any strategies P* and
Q* that satisfy (2) and (3) are called optimal strategies for R and C, respectively. It
can be shown that v is the largest guaranteed expectation that R can obtain, irrespec-
tive of C’s play. Surprisingly, v is also the smallest expectation that C can allow for R,
irrespective of R’s play.
Essentially, the fundamental theorem states that every matrix game has optimal
strategies for R and C and a unique game value v. Finding these optimal strategies
and the corresponding value of a matrix game is called solving the game. The triplet
1v, P*, Q*2 is called a solution of the game.
An immediate consequence of the fundamental theorem is the fact that the ex-
pected value of the game for R, when both R and C use optimal strategies, is v. This
is why we call v the value of the game. We now state this as a theorem (the proof is
left to Problem 42, Exercises 10.2).
THEOREM 3 Expected Value for Optimal Strategies
E1P*, Q*2 = P*MQ* = v
Solution to a 2 : 2 Matrix Game

Now that we know what a solution to a matrix game is, how do we find it? For a
2 * 2 matrix game that is not strictly determined (we already know how to solve
a strictly determined matrix game), there exists a set of formulas for the solution in
terms of the payoff values in the matrix for the game. Although such simple formu-
las do not exist for the solution of m * n matrix games in general, in Sections 10.3
and 10.4 we show that the problem of solving a matrix game can be converted to an
equivalent linear programming problem. We then use the tools developed in Chapter 5
to obtain a solution to the matrix game.
Theorem 4 gives the solution of any 2 * 2 nonstrictly determined matrix game.
The formulas in Theorem 4 can be derived by mimicking the solution of the two-
finger Morra game on pages G9–G11 (see Problems 47 and 48, Exercises 10.2). To
prove Theorem 4, it is sufficient to show that inequalities (2) and (3) in Theorem 2
hold (see Problem 43, Exercises 10.2).
†
Since Theorem 2 applies to every m * n matrix game M, it applies to strictly determined games too. For
the latter, v is a saddle value, and P* and Q* are pure strategies.
THEOREM 4 Solution to a 2 : 2 Nonstrictly Determined Matrix Game

For the nonstrictly determined game
a b
M = c d
c d
the optimal strategies P* and Q* and the value of the game are given by
d - b
d - c a - b q1* D
P* = 3p1* p2*4 = c d Q* = c d = D T
D D q2* a - c
D
ad - bc
v =
D
where D = 1a + d2 - 1b + c2.
Note: Under the assumption that M is not strictly determined, it can be shown that
D = 1a + d2 - 1b + c2 will never be 0.
EXAMPLE 1 Solving a 2 : 2 Nonstrictly Determined Matrix Game Solve the two-finger

Morra game introduced at the beginning of this section.
a b 2 -3
SOLUTION M = c d = c d
c d -3 4
We first compute D: D = 1a + d2 - 1b + c2 = 6 - 1- 62 = 12
Then the solution of the game is
d - c a - b 7 5
P* = c d = c d
D D 12 12
d - b 7
D 12
Q* = D T = D T
a - c 5
D 12
ad - bc 8 - 9 1
v = = = -
D 12 12
Note that the solution agrees with our earlier results: R’s optimal strategy is
7 5
P* = c d
12 12
That is, using a random process (such as colored marbles in an urn or a spinner), R
7 5
should choose row 1 with a probability of 12 and row 2 with a probability of 12 . The
optimal strategy for C is
7
Q* = c 12
5 d
12
7
Using a random process, C should choose column 1 with a probability of 12 and
5
column 2 with a probability of 12. If optimal strategies are used by both R and C,
1
the expected value of the game for R is - 12 . In the long run, R’s average loss will
1 1
be v = - 12 of a dollar per game, while C’s average gain per game will be 12 of a
1
dollar. The game is not fair since v = - 12 ∙ 0. It would be better to be player C.
This is rather surprising since the game matrix M appears to be fair.
Matched Problem 1 Solve the following version of the two-finger Morra

game (which is equivalent to the penny-matching game in Section 10.1):
1 -1
M = c d
-1 1
a b
Let M = c d.
c d
(A) Show that if the row minima belong to the same column, at least one of them is
a saddle value.
(B) Show that if the column maxima belong to the same row, at least one of them is
a saddle value.
(C) Show that if 1a + d2 - 1b + c2 = 0, then M has a saddle value (that is, M is
strictly determined).
(D) Explain why part (C) implies that the denominator D in Theorem 4 will never be 0.
Recessive Rows and Columns

Some strictly determined and nonstrictly determined higher-dimensional matrix
games may be reduced to lower-dimensional games because intelligent players
would never play certain rows or columns. Consider the following matrix game:
-3 0 1
M = C 3 -1 2 S
-2 1 1
Note that M has no saddle value so the game is not strictly determined. Now, from R’s
point of view, there is a row that should never be played. Can you determine which
one? If we compare row 1 with row 3, we find that each element of row 3 is greater
than or equal to the corresponding element of row 1 (that is, - 2 Ú - 3, 1 Ú 0, and
1 Ú 1). Consequently, there would never be an advantage for R to play row 1, so we
delete it to obtain
3 -1 2
M1 = c d
-2 1 1
From C’s point of view, there is a column that should never be played. Can you find
it? If we compare column 3 with column 2, we find that each element of column 2 is less
than or equal to the corresponding element of column 3 (that is, - 1 … 2 and 1 … 1).
There would never be an advantage for C to play column 3, so we delete it to obtain
3 -1
M2 = c d
-2 1
The game is now 2 * 2 and can be solved using the formulas in Theorem 4. The con-
cept of recessive rows and recessive columns is helpful in describing such a reduction.
DEFINITION Recessive Rows and Columns

A row of a payoff matrix is a recessive row if there exists another row, called a
dominant row, such that each element of the dominant row is greater than or
equal to 1 Ú 2 the corresponding element of the recessive row.
A column of a payoff matrix is a recessive column if there exists another col-
umn, called a dominant column, such that each element of the dominant column
is less than or equal to 1 … 2 the corresponding element of the recessive column.
Note that the expected order relation, a Ú b, for an element a of

! CAUTION a dominant row and the corresponding element b of a recessive
row, is reversed and becomes c … d for an element c of a dominant column and the
corresponding element d of a recessive column.
CONCEPTUAL INSIGHT
If r and s are two distinct real numbers, then either r 7 s or r 6 s (one of them
must be greater than the other). A matrix may have two rows (or columns), how-
ever, neither of which is recessive or dominant. Consider, for example,
3 -1
M2 = c d
-2 1
The n * n identity matrix is an example of an n * n matrix that has no recessive
row and no recessive column.
Because player R would never choose a recessive row and player C would never
choose a recessive column, a recessive row or column may be deleted from a game
matrix without affecting the optimal strategy of either player. Suppose that a reces-
sive row or column has been deleted. If the resulting matrix also has a recessive row
or column, then that row or column may also be deleted. By repeating this procedure,
the original matrix can be reduced to one that has no recessive row and no recessive
column without affecting the optimal strategy of either player.
EXAMPLE 2 Deleting Recessive Rows and Columns Solve the matrix game:
-1 -3 1 0
M = C 0 -2 4 0S
3 1 -3 2
SOLUTION
Step 1 Is M strictly determined? No. (If the answer were yes, we would find a sad-
dle value and corresponding pure strategies for R and C, and then we would
be finished.)
Step 2 Delete recessive rows and columns, if present.
Row 2 dominates Column 2 dominates
row 1. columns 1 and 4.
-1 -3 1 0
0 -2 4 0
C 0 -2 4 0S S c d
3 1 -3 2
3 1 -3 2
-2 4
Sc d
1 -3
Step 3 Solve the remaining matrix game:
a b -2 4
M1 = c d = c d
c d 1 -3
D = 1a + d2 - 1b + c2 = 1 - 52 - 152 = - 10
d - c a - b -4 -6 2 3
P* = c d = c d = c d
D D - 10 - 10 5 5
d - b -7 7
D - 10 10
Q* = D T = D T = D T
a - c -3 3
D - 10 10
ad - bc 6 - 4 2 1
v = = = = -
D -10 -10 5
The game is not fair and favors C. Note that in terms of the original game M,
we would say that R’s optimal strategy is
2 3
P* = c 0 d
5 5
which means that player R, using a random process, should choose row 1 with a
probability of 0, row 2 with a probability of 25, and row 3 with a probability of 35. The
optimal strategy for C is
0
7
10
Q* = D3T
10
0
Player C, using a random process, should choose column 1 with a probability of 0,
7 3
column 2 with a probability of 10 , column 3 with a probability of 10 , and column 4
with a probability of 0.
If both players follow their optimal strategies, the expected value of the game
for R is - 15 and for C is 15.
Matched Problem 2 Solve the matrix game:
-1 -1 2
M = C 3 2 -4 S
-1 -1 3
Even though dominance is applicable to strictly determined games, its primary

use is in the reduction of mixed strategy games. Mixed strategy games are much
easier to solve if they can be reduced first.

(A) Using Theorem 4, give conditions on a, b, c, and d to guarantee that the non-
strictly determined matrix game
a b
M = c d
c d
is fair.
(B) Construct the matrix of payoffs for a two-finger Morra game that is nonstrictly
determined and fair.
(C) How many such matrices are there? Explain.
Exercises 10.2
Skills Warm-up Exercises 2 - 3 .5 2 - 3 .5
5. [.5 .5]c dc d 6. [.4 .6]c dc d
W In Problems 1–8, calculate the matrix product. (If necessary, -1 4 .5 -1 4 .5
review Secion 3.4).
2 -3 0 2 -3 1 2 - 3 .7 2 - 3 .7
1. [1 0]c dc d 2. [0 1]c dc d 7. [.4 .6]c dc d 8. [.5 .5]c dc d
-1 4 1 -1 4 0 -1 4 .3 -1 4 .3
2 -3 0 2 -3 1
3. [0 1]c dc d 4. [1 0]c dc d
-1 4 1 -1 4 0
A In Problems 9–18, which rows and columns of the game matrix 37. If the first-column entries of a 2 * 2 payoff matrix are equal,
are recessive? then the game is strictly determined.
38. If a matrix game is fair, then both players have optimal strate-
2 -1 1 3
9. c d 10. c d gies that are pure.
3 5 -2 0
C 39. You (R) and a friend (C) are playing the following matrix
-3 5 3 2 -4 3 game, where the entries indicate your winnings from C in dol-
11. c d 12. c d
-1 0 -1 1 2 -5 lars. In order to encourage your friend to play, since you cannot
lose as the matrix is written, you pay her $3 before each game.
-3 5 2 -5
C
13. C 0 4S 14. C 0 4S 0 2 1 0
2 -1 -1 3 4 3 5 4
R D T
-2 0 -1 2 2 3 0 2 6 1
15. C 0 -5 1S 16. C 4 -1 -1 S 0 1 0 3
3 -2 1 3 0 1 (A) If selection of a particular row or a particular column is
0 1 0 2 0 0 made at random, what is your expected value?
17. C 0 0 1S 18. C 0 1 0S (B) If you make whatever row choice you wish and your oppo-
1 0 0 0 0 3 nent continues to select a column at random, what is your
expected value, assuming that you optimize your choice?
Solve the matrix games in Problems 19–32, indicating optimal
(C) If you both make your own choices, what is your expect-
strategies P* and Q* for R and C, respectively, and the value v
ed value, assuming that you both optimize your choices?
of the game. (Both strictly and nonstrictly determined games are
included, so check for this first.) 40. You (R) and a friend (C) are playing the following matrix
game, where the entries indicate your winnings from C in
-2 1 2 3 dollars. To encourage your friend to play, you pay her $4
19. c d 20. c d before each game. The jack, queen, king, and ace from the
1 0 5 -1
hearts, spades, and diamonds are taken from a standard
1 3 -1 -2 deck of cards. The game is based on a random draw of a
21. c d 22. c d
0 1 0 -3 single card from these 12 cards. The play is indicated at the
top and side of the matrix.
1 -2 3 - 15
23. c d 24. c d C
-2 4 -2 10
H S D
4 3 -2 4 1 2 3 0
B 25. c d 26. c d J 1 0 6
1 5 4 2 5 -5 0 3 Q 2 2 1
R D T
3 -2 0 -3 K 4 4 7
3 4 3 -1 A 1 2 6
27. ≥ ¥ 28. ≥ ¥
5 -1 2 -2 (A) If you select a row by drawing a single card and your
3 2 1 2 friend selects a column by drawing a single card (after
5 3 4 4 3 -1 replacement), what is your expected value of the game?
29. C 1 -2 0S 30. C 3 0 -2 S (B) If your opponent chooses an optimal strategy (ignoring
7 0 -3 -5 -9 3 the cards) and you make your row choice by drawing a
card, what is your expected value?
3 0 -1 1 2 -3
31. C 2 -1 -3 S 32. C 2 0 -1 S (C) If you both disregard the cards and make your own
-6 3 2 1 5 -2 choices, what is your expected value, assuming that you
both choose optimal strategies?
In Problems 33–38, discuss the validity of each statement. If the 41. For
statement is always true, explain why. If not, give a counterexample.
a b
M = c d
33. If a matrix game is strictly determined, then both players c d
have optimal strategies that are pure. q
P = [p1 p2] Q = c 1d
34. If both players of a matrix game have optimal strategies that q2
are mixed, then the game is nonstrictly determined. show that PMQ = E1P, Q2.
35. If a payoff matrix has a row consisting of all 0’s, then that 42. Using the fundamental theorem of game theory, prove that
row is recessive.
P* MQ* = v
36. Every payoff matrix either has a recessive row or a recessive
column.
43. Show that the solution formulas (Theorem 4) for a 2 * 2 (A) Find optimal strategies for bank R and bank C. What is
nonstrictly determined matrix game meet the conditions for a the value of the game?
solution stated in Theorem 2.
(B) What is the expected value of the game for R if bank R
44. Show that if a 2 * 2 matrix game has a saddle value, then always chooses TV and bank C uses its optimal strategy?
either one row is recessive or one column is recessive.
(C) What is the expected value of the game for R if bank
45. Explain how to construct a 2 * 2 matrix game M for which C always chooses radio and bank R uses its optimal
the optimal strategies are strategy?
.3 (D) What is the expected value of the game for R if both
P* = [.9 .1] and Q* = c d
.7 banks always use the newspaper?
46. Explain how to construct a 2 * 2 matrix game M for which
50. Viewer ratings. A city has two competitive television
the optimal strategies are
stations, station R and station C. Every month, each station
.8 makes exactly one choice for the Thursday 8–9 p.m. time slot
P* = [.6 .4] and Q* = c d
.2 from the program categories shown in the following matrix.
In Problems 47 and 48, derive the formulas of Theorem 4 for the Each matrix entry is an average viewer index rating gain (or
solution of any 2 * 2 nonstrictly determined matrix game by loss) established by a rating firm using data collected over the
rewriting and analyzing past 5 years. (Any gain for station R is a loss for station C,
and vice versa.)
E1P, Q2 = a p1q1 + bp1q2 + c p2q1 + dp2q2 (4)
C
(See the solution of the two-finger Morra game on pages G9–G11.) Nature Talk Sports
47. (A) Let p2 = 1 - p1 and q2 = 1 - q1 and simplify (4) to films shows events Movies
show that
Travel 0 2 -1 0
E1P, Q2 = 3Dp1 - 1d - c24q1 + 1b - d2p1 + d News 1 0 -1 -2
R D T
where D = 1a + d2 - 1b + c2. Sitcoms 2 3 -1 1
(B) Show that if p1 is chosen so that Dp1 - 1d - c2 = 0, Soaps 1 -2 0 0
ad - bc
then v = , regardless of the value of p1. (A) Find the optimal strategies for station R and station C.
D
48. (A) Let p2 = 1 - p1 and q2 = 1 - q1 and simplify (4) to What is the value of the game?
show that (B) What is the expected value of the game for R if station
E1P, Q2 = 3Dq1 - 1d - b24p1 + 1c - d2q1 + d R always chooses travel and station C uses its optimal
strategy?
where D = 1a + d2 - 1b + c2.
(B) Show that if q1 is chosen so that Dq1 - 1d - b2 = 0, (C) What is the expected value of the game for R if station
C always chooses movies and station R uses its optimal
ad - bc strategy?
then v = , regardless of the value of p1.
D
(D) What is the expected value of the game for R if station
R always chooses sitcoms and station C always chooses
sports events?
Applications 51. Investment. You have inherited $10,000 just prior to a
presidential election and wish to invest it in solar energy and
49. Bank promotion. A town has only two banks, bank R and oil stocks. An investment advisor provides you with a payoff
bank C, and both compete equally for the town’s business. matrix that indicates your probable 4-year gains, depending
Every week, each bank decides on the use of one, and only on which party comes into office. How should you invest
one, of the following means of promotion: TV, radio, newspa- your money so that you would have the largest expected gain
per, and mail. A market research firm provided the following irrespective of how the election turns out?
payoff matrix, which indicates the percentage of market gain
Player C (fate)
or loss for each choice of action by R and by C (we assume
Republican Democrat
that any gain by R is a loss by C, and vice versa):
Player R Solar energy $1,000 $4,000
C (you) Oil c d
$5,000 $3,000
TV Radio Paper Mail
TV 0 -1 -1 0 Note: For a one-time play (investment), you would split your
Radio 1 2 -1 -1 investment proportional to the entries in your optimal strategy
R
Paper D 0 -1 0 1
T matrix. Assume that fate is a very clever player. Then if fate
Mail deviates from its optimal strategy, you know you will not do
-1 -1 -1 0
any worse than the value of the game, and you may do better.
SECTION 10.3 Games: A Geometric Approach G19
52. Corporate farming. A corporate farm grows both wheat and “wet” for many years, and the farm continues to play its
rice. Rice does better in wetter years, and wheat does better optimal strategy?
in normal or drier years. Based on records over the past 20 (C) Answer part (B), replacing “wet” with “normal.”
years and current market prices, a consulting firm provides
the farm managers with the following payoff matrix, where (D) Answer part (B), replacing “wet” with “dry.”
the entries are in millions of dollars:
Weather (fate)
1
Wet Normal Dry
Corporate Wheat 1. P* = 3 12 1
2 4, Q* = c 21 d , v = 0
-2 8 2 2
farm Rice c d
7 3 -3 2. Eliminate recessive rows and columns to obtain
The managers would like to determine the best strategy against 2 -4
the weather’s “best strategy” to destroy them. Then, no matter M1 = c d
-1 3
what the weather does, the farm will do no worse than the
value of the game and may do a lot better. This information 7
3
could be very useful to the company when applying for loans. Then P* = 3 25 5 4, Q* = c 10
3 d
Note: For each year that the payoff matrix holds, the farm 10
can split the planting between wheat and rice proportional to

the size of entries in its optimal strategy matrix. 2 3
For the original game M, P* = c 0 5 5d,
(A) Find the optimal strategies for the farm and the weather,
and the value of the game. 0
7 1
(B) What is the expected value of the game for the farm Q* = C 10 S , v = 5
3
if the weather (fate) chooses to play the pure strategy 10
10.3 Linear Programming and 2 : 2 Games:

In Section 10.2, we said that formula solutions for the more general m * n matrix
game do not exist. There is, however, a systematic way to solve a nonstrictly determined
matrix game without recessive rows or columns. All such games can be converted into
linear programming problems, which can be solved using the techniques in Chapter 5.
We can describe the linear programming process geometrically using a 2 * 2
matrix game, where ideas and procedures do not get too involved. (A brief review of
Section 4.3 may prove helpful.) In the next section, using the simplex method and the
dual problem from Sections 5.2 and 5.3, we develop a general procedure for arbitrary
m * n matrix games.
Consider a nonstrictly determined matrix game with all positive payoffs:
Player C
a b
Player R c d = M a, b, c, d, 7 0 (1)
c d
The all positive payoff requirement is necessary to guarantee that v (the value of the
game) is positive, a fact we will need shortly. The requirement of all positive payoffs
may seem too restrictive. However, it causes little trouble because of the following
theorem, which we state without proof (see Problems 25 and 26, Exercises 10.3):
THEOREM 1 Invariant Optimal Strategies

Optimal strategies of a matrix game do not change if a constant value k is added to each
payoff. If v is the value of the original game, then v + k is the value of the new game.
For example, if we start with the matrix game
2 -3
M = c d
-1 0
then all we have to do is add some number, say, 4, to each payoff to obtain the positive
matrix
6 1
M1 = c d
3 4
We find P*, Q *, and v1 for this new game. Then P*, Q*, and v = v1 - 4 will be the
solution of the original game.
Recall from Section 10.2 that we introduced the idea of expected value when
looking for optimal strategies for R and C. We found that for a matrix game M, the
expected value of the game for R is given by
E1P, Q2 = PMQ (2)
where
q1
P = 3p1 p2 4 is R’s strategy and Q = c d is C’s strategy
q2
Naturally, R is interested in finding a strategy P*, along with the largest possible
number v, such that
E1P*, Q2 = P*MQ Ú v (3)
for any choice of Q by C.
Since inequality (3) is to hold for all Q, it must hold in particular for
1 0
Q1 = c d and Q2 = c d
0 1
Using the positive game matrix (1), and successively substituting Q1 and Q2 into
inequality (3), we obtain the two inequalities
a b 1 a b 0
3p1 p2 4 c dc d Ú v 3p1 p2 4 c dc d Ú v
c d 0 c d 1
Multiplying, we have
ap1 + cp2 Ú v bp1 + dp2 Ú v (4)
These inequalities suggest a linear programming problem, but each involves three
variables, p1, p2, and v, instead of two. We remedy this by dividing through by v.
(The sense of the inequalities will not change since we know that v is positive by the
assumption that each payoff in M is positive.) Inequalities (4) now become
p1 p2 p1 p2
a + c Ú 1 b + d Ú 1 (5)
v v v v
To simplify notation, we introduce two new variables,
p1 p2
x1 = and x2 = where x1 Ú 0, x2 Ú 0 (6)
v v
Inequalities (5) can now be written in the simpler form
ax1 + cx2 Ú 1
x1, x2 Ú 0 (7)
bx1 + dx2 Ú 1
Our problem now is to maximize v subject to the constraints in inequalities (7).
Adding the two equations in (6) produces the following useful result:
p1 p2
x1 + x2 = +
v v
p1 + p2
= Use p1 + p2 = 1. (Why?)
v
1
=
v
So
1 1
x1 + x2 = or v = (8)
v x1 + x2
To maximize v, we minimize 1>v instead (as v gets larger, 1>v gets smaller). We
now have the following linear programming problem:
1
Minimize y = = x1 + x2 (9)
v
subject to ax1 + cx2 Ú 1
bx1 + dx2 Ú 1
x1, x2 Ú 0
Using the methods discussed in Section 4.3, we solve problem (9) geometrically and
then use equations (8) and (6) to find v and P*.
Now we turn to player C, who is naturally interested in finding a strategy Q* and
the smallest possible value v such that
E1P, Q*2 = PMQ* … v (10)
for any choice of P by R.
Since inequality (10) is to hold for all P, it must hold in particular for
P1 = 31 04 and P2 = 30 14
Substituting P1 and P2 into inequality (10) and multiplying, we obtain
aq1 + bq2 … v
(11)
cq1 + dq2 … v
Dividing through by the positive number v and letting
q1 q2
z1 = and z2 = where z1 Ú 0, z2 Ú 0 (12)
v v
we obtain
az1 + bz2 … 1
z1, z2 Ú 0 (13)
cz1 + dz2 … 1
We are interested in minimizing v subject to the constraints (13). As before, we
note that
q1 q2
z1 + z2 = +
v v
q1 + q2
=
v
1
=
v
Therefore,
1 1
z1 + z2 = or v = (14)
v z1 + z2
Minimizing v is the same as maximizing 1>v (decreasing v increases 1>v). So we
have a second linear programming problem:
1
Maximize y = = z1 + z2 (15)
v
subject to az1 + bz2 … 1
cz1 + dz2 … 1
z1, z2 Ú 0
After problem (15) is solved geometrically, we find v and Q* using equations (14)
and (12).
From Section 5.3, you should recognize problem (15) as the dual of problem (9). And
by Theorem 1 in that section, problems (9) and (15) must have the same optimal value.
The v found using problem (9) and the v found using problem (15) must be the same.
The process of converting a 2 * 2 matrix game into a linear programming prob-
lem generalizes to m * n matrix games, which we consider in the next section. We
summarize the preceding results for a 2 * 2 matrix game in the next box for conve-
nient reference.
PROCEDURE 2 : 2 Matrix Games and Linear Programming:

Geometric Approach
Given the nonstrictly determined matrix game
a b
M = c d
c d
q1
find P* = 3p1 p2 4, Q* = c d , and v, as follows:
q2
Step 1 If M is not a positive matrix (one with all entries positive), then convert it
into a positive matrix M1 by adding a suitable positive constant k to each
element. Let M1, the new positive matrix, be represented as follows:
e f e = a + k f = b + k
M1 = c d
g h g = c + k h = d + k
This new matrix game, M1, has the same optimal strategies P* and Q* as M.
However, if v1 is the value of the game M1, then
v = v1 - k
is the value of the original game M.
Step 2 Set up the two corresponding linear programming problems:
(A) Minimize y = x1 + x2 (B) Maximize y = z1 + z2
subject to ex1 + gx2 Ú 1 subject to ez1 + fz2 … 1
fx1 + hx2 Ú 1 gz1 + hz2 … 1
x1, x2 Ú 0 z1, z2 Ú 0
Step 3 Solve each linear programming problem geometrically.
Step 4 Use the solutions in step 3 to find the value v1 for game M1 and the optimal
strategies and value v for the original game M:
1 1 1 1
v1 = = or v1 = =
y x1 + x2 y z1 + z2
q1 vz
P* = 3p1 p2 4 = 3v1x1 v1x2 4 Q* = c d = c 1 1d
q2 v1z2
v = v1 - k
Step 5 A further check of the solution is made by showing that
P*MQ* = v See Theorem 3, Section 10.2.
CONCEPTUAL INSIGHT
In step 2 in the box, note that part (B) is the dual of part (A), and consequently both
must have the same optimal value (Theorem 1, Section 5.3). In the next section, us-
ing the simplex method and properties of the dual, we will see that solving part (B)
will automatically produce the solution for part (A). In this section, we restrict our at-
tention to the geometric approach, so we must solve each part as a separate problem.
EXAMPLE 1 Solving 2 : 2 Matrix Games Using Geometric Methods Solve the following
matrix game by using geometric methods to solve the corresponding linear pro-
gramming problems (see Section 4.3):
-2 4
M = c d
1 -3
SOLUTION
Step 1 Convert M into a positive matrix (one with all entries positive) by adding
4 to each payoff. We denote the modified matrix by M1:
2 8 e f
M1 = c d = c d k = 4
5 1 g h
(A) Minimize y = x1 + x2 (B) Maximize y = z1 + z2

subject to 2x1 + 5x2 Ú 1 subject to 2z1 + 8z2 … 1
8x1 + x2 Ú 1 5z1 + z2 … 1
x1, x2 Ú 0 z1, z2 Ú 0
Step 3 Solve each linear programming problem geometrically:
x2 z2
1.0
0.2
Feasible
region
0.5
0.1
Feasible
region
x1 z1
0 0.5 1.0 0 0.1 0.2
Theorems 1 and 2 in Section 4.3 imply that each problem has a solution that must
occur at a corner point.
(A) Corner Minimize (B) Corner Maximize
Points y = x1 + x2 Points y = z1 + z 2
10, 12 1 10, 02 0
1 192 , 193 2 5
19 1 0, 18 2 1
8
1 12, 02 1
2 1 387 , 383 2 5
19
Min y occurs at 1 15, 0 2 1

5
2 3
x1 = 19 and x2 = 19 Max y occurs at
7 3
z1 = 38 and z2 = 38
Step 4 Use the solutions in step 3 to find the value v1 for the game M1 and the
optimal strategies and value v for the original game M:
1 1 19 1 1 19
(A) v1 = = 2 3 = (B) v1 = = 7 3 =
x1 + x2 19 + 19
5 z1 + z2 38 + 38
5
p1 = v1x1 = #
19 2
5 19 = 2
5
q1 = v1z1 = 19
5
# 7 7
38 = 10
p2 = v1x2 = #
19 3
5 19 = 3
5
q2 = v1z2 = 19
5
# 3 3
38 = 10
Note: v1 found in part (A) should always be the same as v1 found in part (B).
Optimal strategies are the same for both games M and M1. Therefore,
7
3 q1
P* = 3p1 p2 4 = 3 25 54 Q* = c d = c 10
3 d
q2 10
and the value of the original game is

19
v = v1 - k = 5 - 4 = - 15
Step 5 A further check of the solution is provided by showing that
P*MQ* = v See Theorem 3, Section 10.2.
This check is left to the reader.
Matched Problem 1 Solve the following matrix game using geometric linear
programming methods:
2 -4
M = c d
-1 3

Show that
1 -1
M = c d
-3 -2
is a strictly determined matrix game. Nevertheless, apply the geometric procedure
given for nonstrictly determined matrix games to M. Does something go wrong? Do
you obtain the correct optimal strategies? Explain.
Exercises 10.3
A In Problems 1–6, find the smallest integer k Ú 0 such that adding 14. Is there a better way to solve the matrix game in Problem 12
k to each entry of the given matrix produces a matrix with all than the geometric linear programming approach? Explain.
positive payoffs.
15. Explain why the value of a matrix game is positive if all of
the payoffs are positive.
-3 -5 4 -1
1. c d 2. c d
2 -6 -1 2 16. Explain why the value of a matrix game is negative if all of
the payoffs are negative.
0 1 6 8
3. c d 4. c d In Problems 17–20, discuss the validity of each statement. If the
2 0 5 1
statement is always true, explain why. If not, give a counterex-
1 2 -3 -2 ample.
5. c d 6. c d
4 7 -1 -3 17. If all payoffs of a matrix game are zero, then the game
is fair.
In Problems 7–12, solve the matrix games using a geometric
18. If the value of a matrix game is positive, then all payoffs are
linear programming approach.
positive.
2 -3 2 -1 19. If half of the payoffs of a game matrix are positive and half
7. c d 8. c d
-1 2 -2 1 are negative, then the game is fair.
-1 3 4 -6 20. If a matrix game is fair, then some payoffs are positive and
9. c d 10. c d some are negative.
2 -6 -2 3
-2 -1 6 2 In Problems 21–24 remove recessive rows and columns; then

11. c d 12. c d solve using geometric linear programming techniques.
5 6 -1 1
B 13. Is there a better way to solve the matrix game in Problem 11 12 0 -4 15 1 9 -3 10

21. c d 22. c d
than the geometric linear programming approach? Explain. -3 2 1 0 2 -3 1 0
SECTION 10.4 Games: Simplex Method and the Dual Problem G25
Applications
1 -3 -5 5
3 -2 1 -3
23. D T 24. D T
-6 4 -1 -5 Solve the matrix games in Problems 27–30 by using geometric
-8 2 -2 6 linear programming methods.
C 25. (A) Let P and Q be strategies for the 2 * 2 matrix game M. 27. Bank promotion. Problem 49A, Exercises 10.2
Let k be a constant, and let J be the matrix with all 1’s as 28. Viewer ratings. Problem 50A, Exercises 10.2
entries. Show that the matrix product P(kJ) Q equals the
1 * 1 matrix k. 29. Investment. Problem 51, Exercises 10.2
(B) Generalize part (A) to the situation where M and J are 30. Corporate farming. Problem 52A, Exercises 10.2
m * n matrices.
26. Use properties of matrix addition and multiplication to Answer to Matched Problem
deduce from Problem 25 that if P* and Q* are optimal strate- 7
gies for the game M with value v, then they are also optimal 1. P* = 3 25 3
5 4 , Q* = c 10
3 d, v =
1
5
strategies for the game M + kJ with value v + k. 10
10.4 Linear Programming and m : n Games: Simplex Method

and the Dual Problem
In this section, we generalize the process of solving 2 * 2 matrix games to m * n
matrix games. The simplex method, using the dual problem, will play a central role
in the solution process.
The methods outlined in Section 10.3 for converting a 2 * 2 matrix game into
a pair of linear programming problems, each the dual of the other, generalize com-
pletely to m * n matrix games. We restate the procedure for a 2 * 3 game, illustrate
the procedure with an example, and solve the example using simplex methods and the
dual problem. The fact that the two linear programming problems are mutually dual
provides the advantage that the solution of one automatically gives the solution of the
other (see Section 5.3). Based on the analysis of a 2 * 3 game, you should be able to
convert any m * n matrix game into a pair of linear programming problems, one the
dual of the other, and solve them using the methods of Section 5.3.
PROCEDURE 2 : 3 Matrix Games and Linear Programming: Simplex

Method and the Dual Problem
Given the nonstrictly determined matrix game M, free of recessive rows and columns,
r r2 r3
M = c 1 d
s1 s2 s3
q1
find P* = 3p1 p2 4, Q* = C q2 S , and v, as follows:
q3
Step 1 If M is not a positive matrix, convert it into a positive matrix M1 by adding

a suitable positive constant k to each element. Let M1, the new positive
matrix, be represented as follows:
a a2 a3
M1 = c 1 d
b1 b2 b3
If v1 is the value of the game M1, then the value of the original game M is
given by v = v1 - k.
Step 2 Set up the two linear programming problems (the maximization problem is
always the dual of the minimization problem):
(A) Minimize y = x1 + x2
subject to a1x1 + b1x2 Ú 1
a2x1 + b2x2 Ú 1
a3x1 + b3x2 Ú 1
x1, x2 Ú 0
(B) Maximize y = z1 + z2 + z3
subject to a1z1 + a2z2 + a3z3 … 1
b1z1 + b2z2 + b3z3 … 1
z1, z2, z3 Ú 0
Step 3 Solve the maximization problem, part (B), the dual of part (A), using the
simplex method as modified in Section 5.3. [You will automatically get the
solution of the minimization problem, part (A), as well, by following this
process.]
Step 4 Use the solutions in step 3 to find the value v1 for game M1 and the opti-
mal strategies and value v for the original game M:
1 1 1 1
v1 = = or v1 = =
y x1 + x2 y z1 + z2 + z3
q1 v1z1
P* = 3p1 p2 4 = 3v1x1 v1x2 4 Q* = C q2 S = C v1z2 S
q3 v1z3
v = v1 - k
Step 5 A further check of the solution is provided by showing that
P*MQ* = v
CONCEPTUAL INSIGHT
This procedure is easily extended to an m * n matrix. After adding an appropriate
k, we get a matrix
a1 a2 a3 c
b1 b2 b3 c
M1 = ≥ c
1 c2 c3 c ¥ .
f f f f
We minimize y = x1 + x2 + c + xm subject to n equations of the form
a1x1 + b1x2 + c Ú 1. We maximize y = z1 + z2 + c + zn subject to
m equations of the form d1z1 + d2z2 + c + dnzn … 1. We get that v1 = 1y ,
v1z1
P* = 3v1x1 v2x2 c 4, Q* = C v2z2 S , and v = v1 - k.
f
EXAMPLE 1 Investment Analysis An investor wishes to invest $10,000 in bonds and gold.
He knows that the return on the investments will be affected by changes in interest
rates. After some analysis, he estimates that the return (in thousands of dollars) at
the end of a year will be as indicated in the following payoff matrix:
Change in interest rates (fate)
0% + 1% - 3%
Bonds 1 -1 6
c d
Gold -1 2 -3
(A) We assume that fate is a very clever player and will play to reduce the investor’s
return as much as possible. Find optimal strategies for both the investor and for
“fate.” What is the value of the game?
(B) Find the expected values of the game if the investor continues with his optimal
strategy and fate “switches” to the following pure strategies: (1) Play only 0%
change; (2) Play only + 1% change; (3) Play only - 3% change.
SOLUTION
1 -1 6
(A) Let M = c d
-1 2 -3
Step 1 Convert M into a positive matrix M1 by adding 4 to each entry in M:
5 3 10
M1 = c d k = 4
3 6 1
(A) Minimize y = x1 + x2
3x1 + 6x2 Ú 1
10x1 + x2 Ú 1
x1, x2 Ú 0
(B) Maximize y = z1 + z2 + z3
subject to 5z1 + 3z2 + 10z3 … 1
3z1 + 6z2 + z3 … 1
z1, z2, z3 Ú 0
Step 3 Solve the maximization problem, part (B), the dual of part (A), using the
simplex method as modified in Section 5.3. We introduce slack variables
x1 and x2 to obtain
5z1 + 3z2 + 10z3 + x1 = 1
3z1 + 6z2 + z3 + x2 = 1
- z1 - z2 - z3 + y = 0
Write the simplex tableau and identify the first pivot element:
z1 z2 z3 x1 x2 y
1
5 3 10 1 0 0 1 5 1minimum2
C 3 6 1 0 1 0 3 1S 1
3
-1 -1 -1 0 0 1 0
Since the most negative indicator in the third row appears in each of the first three
columns, we can choose any of these as a pivot column. We choose the first
column and then find that the first row is the pivot row. Now we use row opera-
tions to pivot on 5:
z1 z2 z x1 x2 y
1
x1 5 3 10 1 0 0 1 5 R1 S R1
x2 C 3 6 1 0 1 0 3 1S
y -1 -1 -1 0 0 1 0
3 1
1 5 2 5 0 0 15
∼C 3 6 1 0 1 0 3 1S 1 - 32R1 + R2 S R2
-1 -1 -1 0 0 1 0 R1 + R3 S R3
3
z1 1 5 2 15 0 0 1
5
1
5 , 35 = 13
0 3
2
∼x2 C 0
21
5 - 5 - 35 1 2
5S 5 , 21 2
5 = 21 1minimum2
y 0 - 25 1 15 0 1 1
5
Now select the next pivot element:

z1 z2 z3 x1 x2 y
3 1 1
z1 1 5 2 5 0 0 5
x2 C 0 21
5 -5 - 35 1 0 3 2
5 S 5
21 R2 S R2
y 0 - 25 1 1
5 0 1 1
5
3 1 1
1 5 2 5 0 0 5 1 - 35 2R2 + R1 S R1
∼C 0 1 - 25
21 - 17 5
21 0 3 2
21 S
0 - 25 1 1
5 0 1 1
5
2
R2 + R3 S R3
5
19 2
z1 1 0 7 7 - 17 0 1
7
∼z2 C 0 1 - 25
21 - 17 5
21 0 3 2
21 S
11 1 2 5
y 0 0 21 7 21 1 21
5
Max y = z1 + z2 + z3 = 21 occurs at z1 = 17, z2 = 21
2
, z3 = 0, x1 = 0, x2 = 0.
The solution to the minimization problem, part (A), can be read from the bottom
row of the final simplex tableau for the preceding dual problem. From the row
x1 x2
30 0 11
21
1
7
2
21 1 5
21 4
we conclude that the solution to the minimization problem, part (A), is Min
5
y = x1 + x2 = 21 at x1 = 17, x2 = 21
2
.
Step 4 Use the solutions in step 3 to find the value v1 for the game M1 and the opti-
mal strategies and value v for the original game M:
1 1 21
v1 = = 5 =
y 21
5
P* = 3p1 p2 4 = 3v1x1 v1x2 4 = 3 35 2
5 4
3
5
q1 v1z1
Q* = C q2 S = C v1z2 S = D 2 T
5
q3 v1z3
0
21 1
v = v1 - k = 5 - 4 = 5
Step 5 A further check is made by showing that

P*MQ* = v
This we leave to the reader.
Conclusion: If the investor splits the $10,000 proportional to the numbers in his op-
timal strategy, $6,000 1 35 of $10,000 2 in bonds and $4,000 1 25 of $10,000 2 in gold,
then no matter which strategy fate chooses for interest rate changes (as long as the
payoff matrix remains unchanged), the investor will be guaranteed a return of $200
1 15 of $1,000 2 . If fate plays other than the optimal column strategy, the investor can
do no worse than a $200 return and may do quite a bit better.
(B) Recall that the expected value of the game M using strategies P and Q is
E1P, Q2 = PMQ
If the investor continues to use his optimal strategy P* and fate “switches” to the
pure strategies (1) 0% change, (2) + 1% change, and (3) - 3% change, the expected
values are
P* M Q
1
3 2 1 -1 6 1
112 c dc d C 0 S = 1or $2002
5 5 -1 2 -3 5
0
P* M Q
0
3 2 1 -1 6 1
122 c dc d C 1 S = 1or $2002
5 5 -1 2 -3 5
0
P* M Q
0
3 2 1 -1 6 12
132 c dc d C0S = 1or $2,4002
5 5 -1 2 -3 5
1
Notice that in none of these three cases did the investor earn less than $200, and in
the third case, he earned quite a bit more.
Matched Problem 1 Suppose that the investor in Example 1 wishes to in-

vest $10,000 in long- and short-term bonds, as well as in gold, and he is con-
cerned about inflation. After some analysis, he estimates that the return (in
thousands of dollars) at the end of a year will be as indicated in the following
payoff matrix:
Inflation (fate)
Up 3% Down 3%
Gold 3 -3
Long-term bonds C -3 2S
Short-term bonds -1 1
Again, assume that fate is a very good player that will attempt to reduce the investor’s
return as much as possible. Find the optimal strategies for both the investor and for
fate. What is the value of the game?

Outline a procedure for solving the 4 * 5 matrix game
2 -2 -1 6 -1
-3 -6 4 -1 -7
M = D T
5 3 6 0 4
-4 2 7 3 -5
without actually solving the game.
Exercises 10.4
In Problems 1–4, solve each matrix game.
Applications
1 -1
1 4 0
A 1. c d 2. C 2 -2 S 11. Headphone sales. A department store chain is about to order
0 -1 2
0 1 deluxe, standard, and discount headphones for next year’s
inventory. The state of the nation’s economy (fate) during the
0 1 -2 1 2 0 year will be an important factor on sales for that year. Records
B 3. C - 1 0 3S 4. C 0 1 2S over the past 5 years indicate that if the economy is up, the
2 -3 0 2 0 1 company will net 3, 1, and 0 million dollars, respectively, on
sales of deluxe, standard, and discount models. If the economy
is down, the company will net - 1, 1, and 3 million dollars,
In Problems 5–8, outline a procedure for solving the matrix game, respectively, on sales of deluxe, standard, and discount models.
then solve it. (A) Set up a payoff matrix for this problem.
4 2 2 1 -5 -6 -7 4 (B) Find optimal strategies for both the company and fate
5. C 0 1 -1 3S 6. C - 5 3 2 3S (the economy). What is the value of the game?
-2 -1 -3 2 -4 -2 -6 7 (C) How should the company’s budget be allocated to each
grade of headphone to maximize the company’s return ir-
-2 -1 3 -1 2 -5 -3 -1 respective of what the economy does the following year?
1 2 4 0 0 1 2 -2 12. Tour agency. A tour agency organizes standard and luxury
7. D T 8. D T
-1 1 -1 1 -1 0 1 3 tours for the following year. Once the agency has committed
0 1 -1 2 2 -3 -2 0 to these tours, the schedule cannot be changed. The state of
the economy during the following year has a direct effect on
C 9. Scissors, paper, stone game. This game is well known in tour sales. From past records the agency has established the
many parts of the world. Two players simultaneously present following payoff matrix (in millions of dollars):
a hand in one of three positions: an open hand (paper), a Economy (fate)
closed fist (stone), or two open fingers (scissors). The payoff Down No change Up
is 1 unit according to the rule “Paper covers stone, stone Standard 1 2 0
breaks scissors, and scissors cut paper.” If both players pres- c d
Luxury 0 1 3
ent the same form, the payoff is 0.
(A) Set up the payoff matrix for the game. (A) Find optimal strategies for both the agency and fate (the
economy). What is the value of the game?
(B) Solve the game using the simplex method discussed in
this section. (Remove any recessive rows and columns, (B) What proportion of each type of tour should be arranged for
if present, before you start the simplex method.) in advance in order for the agency to maximize its return
10. Player R has a $2, a $5, and a $10 bill. Player C has a irrespective of what the economy does the following year?
$1, a $5, and a $10 bill. Each player selects and shows (C) What is the expected value of the game to the agency if
(simultaneously) one of his or her three bills. If the total they organize only luxury tours and fate plays the strat-
value of the two bills shown is even, R wins C’s bill; if egy “down”? If the agency plays its optimal strategy and
the value is odd, C wins R’s bill. (Which player would fate plays the strategy “no change”? Discuss these and
you rather be?) other possible scenarios.
(A) Set up the payoff matrix for the game.
Answer to Matched Problem
(B) Solve the game using the simplex method discussed in 1
this section. (Remove any recessive rows and columns, if 1. P* = 3 14 0 3
4 4 , Q* = c 21 d , v = 0
present, before you start the simplex method.) 2
Summary and Review G31

10.1 Strictly Determined Matrix Games EXAMPLES
• Any m * n matrix A may be considered a two-person zero-sum matrix game in which player R
chooses any of the rows and player C chooses any of the columns. We call the matrix A a payoff matrix
or game matrix, and we refer to the entries of A as payoff values. Positive payoff values indicate a win
for R, and negative values a win for C.
• We assume the fundamental principle of game theory, namely, that
1. A matrix game is played repeatedly.
2. Player R tries to maximize winnings.
3. Player C tries to minimize losses.
• A saddle value is a payoff value that is simultaneously a row minimum and column maximum. A proce-
dure for locating saddle values is given on page G4.
• If a payoff matrix has saddle values x and y, then x = y (Theorem 1, page G4).
• A matrix game is said to be strictly determined if it has a saddle value.
• In a strictly determined game, optimal strategies are as follows: Ex. 1, p. G4
R should choose any row containing a saddle value.
C should choose any column containing a saddle value.
• A saddle value is called the value of a strictly determined game.
• A game is fair if its value is 0.
• If a matrix game has no saddle value, then it is said to be nonstrictly determined. Ex. 2, p. G6
• Given the game matrix
a b
M = c d
c d
R’s strategy is a probability row matrix
P = 3p1 p2 4 p1 Ú 0, p2 Ú 0, p1 + p2 = 1
• C’s strategy is a probability column matrix

q
Q = c 1 d q1 Ú 0, q2 Ú 0, q1 + q2 = 1
q2
If one of the elements in P (or Q) is 1 and the other is 0, the strategy is called a pure strategy; otherwise
it is called a mixed strategy.
• The expected value of a matrix game for R, denoted as E(P, Q) is the 1 * 1 matrix product PMQ (Theo-
rem 1, page G11).
• The fundamental theorem of game theory guarantees that R and C have optimal strategies in any
m * n matrix game (Theorem 2, page G12).
• If P* and Q* are optimal strategies for R and C, then the expected value of the game for R is
E(P*, Q*) = P*MQ* = v (Theorem 3, page G12).
• The number v is the value of the game. If v = 0 the game is said to be fair.
• Finding the optimal strategies P* and Q* and the corresponding value v is called solving the game. The
triplet 1v, P*, Q*2 is called a solution of the game.
• For any 2 * 2 nonstrictly determined matrix game, there are formulas (in terms of the entries of M) for Ex. 1, p. G13
P*, Q*, and v that solve the game (Theorem 4, page G13).
• A row of a payoff matrix is a recessive row if there exists another row, called a dominant row, such
that each element of the dominant row is greater than or equal to 1 Ú 2 the corresponding element of the
recessive row. A column of a payoff matrix is a recessive column if there exists another column, called a
dominant column, such that each element of the dominant column is less than or equal to 1 … 2 the cor-
responding element of the recessive column.
• Because player R would never choose a recessive row, and player C would never choose a recessive Ex. 2, p. G15
column, a recessive row or column may be deleted from a payoff matrix without affecting the optimal
strategy of either player.
10.3 Linear Programming and 2 * 2 Games: A Geometric Approach
• Optimal strategies of a matrix game do not change if a constant value k is added to each payoff. If v is
the value of the original game, then v + k is the value of the new game (Theorem 1, page G19).
• Any 2 * 2 nonstrictly determined matrix game can be solved by converting the matrix game to two linear Ex. 1, p. G23
programming problems, each of which can be solved geometrically. See the procedure on page G22.
10.4 Linear Programming and m * n Games: Simplex Method and the Dual Problem
• Any nonstrictly determined m * n matrix game can be solved by converting it to a pair of linear pro- Ex. 1, p. G27
gramming problems, one the dual of the other, which can be solved by the simplex method of Section
5.3. See the procedure on pages G25 and G26.
Review Exercises
Work through all the problems in this chapter review and check 13. Delete as many recessive rows and columns as possible, then
your answers in the back of the book. Answers to all review prob- write the reduced matrix game:
where each type of problem is discussed. Where weaknesses show -2 3 5
up, review appropriate sections in the text. C -1 -3 0S
0 -1 1
In Problems 1 and 2, is the matrix game strictly determined?
5 -3 -1 4 Problems 14–17 refer to the matrix game:

1. c d 2. c d
-4 2 -2 1
-2 1
In Problems 3–8, determine the value v of the matrix game. Is the M = c d
0 -1
game fair?
14. Solve M using formulas from Section 10.2.
5 -2 7 3 15. Write the two linear programming problems corresponding to
3. c d 4. c d
3 0 -9 -1 M after adding 3 to each payoff.
16. Solve the matrix game M using linear programming and a
9 -3 6 - 10
5. c d 6. c d geometric approach.
-5 2 -3 5
17. Solve the matrix game M using linear programming and the
-8 7 -3 4 simplex method.
7. c d 8. c d
-5 4 5 -2 In Problems 18–21, discuss the validity of each statement. If the
statement is always true, explain why. If not, give a counterex-
In Problems 9–12, for each matrix game that is strictly deter- ample.
mined (if it is not strictly determined, say so),
18. If a matrix game is fair, then it is strictly determined.
(A) Locate the saddle values. 19. If a game matrix has a saddle value equal to 0, then the game
(B) Find optimal strategies for R and C. is fair.
(C) Find the value of the game. 20. A game matrix can have at most one recessive row.
-2 2 -2 4 21. If all payoffs of a matrix game are negative, then the value of
9. c d 10. c d
-3 7 3 -6 the game is negative.
-3 -1 5 -8 In Problems 22–26, solve each matrix game (first check for saddle
1 -2 3 values, recessive rows, and recessive columns).
1 0 0 2
11. D T 12. C - 1 2 0S
3 0 1 0 -1 2 8 -1 5 -3 7
3 0 -4
6 -2 -4 2 22. C 0 2 -4 S 23. C - 4 -3 2 -2 S
0 1 3 3 0 2 1
Review Exercises G33
2 6 -4 -7 matrix has been determined, where the entries are in tens of

0 3 -1 25. C 4 7 -3 -5 S thousands of dollars.
24. c d
-1 -2 1 3 3 9 8 Weather (fate)
Wet Dry
Corn 8 4
-1 1 -2 Farmer c d
Soybeans 2 10
26. C 0 -2 2S
-3 2 -1 Use linear programming and the simplex method to find
optimal strategies for the farmer and the weather.
27. Does every strictly determined 2 * 2 matrix game have a 32. Agriculture. Refer to Problem 31. Use formulas from
recessive row or column? Explain. Section 10.2 to find the expected value of the game to the
28. Does every strictly determined 3 * 3 matrix game have a farmer. What is the expected value of the game to the farmer
recessive row or column? Explain. if the weather plays the strategy “dry” for many years and the
farmer always plants soybeans?
29. Finger game. Consider the following finger game be-
tween Ron (rows) and Cathy (columns): Each points either 33. Advertising. A small town has two competing grocery
1 or 2 fingers at the other. If they match, Ron pays Cathy stores, store R and store C. Every week each store decides to
$2. If Ron points 1 finger and Cathy points 2, Cathy pays advertise its specials using either a newspaper ad or a mail-
Ron $3. If Ron points 2 fingers and Cathy points 1, Cathy ing. The following payoff matrix indicates the percentage of
pays Ron $1. market gain or loss for each choice of action by store R and
store C.
(A) Set up a payoff matrix for this game.
(B) Use formulas from Section 10.2 to find the optimal C
strategies for Ron and for Cathy. Paper Mail
30. Refer to Problem 29. Use linear programming and a geomet- Paper 1 -6
ric approach to find the expected value of the game for Ron. R c d
Mail -5 4
What is the expected value for Cathy?
Use linear programming and a geometric approach to find
optimal strategies for store R and store C.
Applications 34. Advertising. Refer to Problem 33. Use linear programming

and the simplex method to find the expected value of the
31. Agriculture. A farmer decides each spring whether to plant game for store R. If store R plays its optimal strategy and
corn or soybeans. Corn is the better crop under wet condi- store C always places a newspaper ad, what is the expected
tions, soybeans under dry conditions. The following payoff value of the game for store C?
Appendix
A
A.1 Linear Equations and
Inequalities
A.2 Graphs and Lines
Linear Equations
and Graphs
Introduction
How far will a glacier advance or retreat in the next ten years? The key to answer-
ing such a question, and to making other climate-related predictions, is mathemat-
A.3 Linear Regression ical modeling. In Appendix A, we study one of the simplest mathematical models,
a linear equation. We introduce a technique called linear regression to construct
mathematical models from numerical data. We use mathematical models to pre-
dict average annual temperature and average annual precipitation (see Problems
23 and 24 in Section A.3), the atmospheric concentration of carbon dioxide, the
consumption of fossil fuels, and many other quantities in business, economics, life
sciences, and social sciences.
530
SECTION A.1 Linear Equations and Inequalities 531
A.1 Linear Equations and Inequalities

■■ Linear Equations The equation
■■ Linear Inequalities x
3 - 21x + 32 = - 5
■■ Applications 3
and the inequality
x
+ 213x - 12 Ú 5
2
are both first degree in one variable. In general, a first-degree, or linear, equation in
one variable is any equation that can be written in the form
Standard form: ax ∙ b ∙ 0 a 3 0 (1)
If the equality symbol, =, in (1) is replaced by 6 , 7 , …, or Ú, the resulting ex-
pression is called a first-degree, or linear, inequality.
A solution of an equation (or inequality) involving a single variable is a number
that, when substituted for the variable, makes the equation (or inequality) true. The
set of all solutions is called the solution set. To solve an equation (or inequality)
means to find its solution set.
Knowing what is meant by the solution set is one thing; finding it is another. We
start by recalling the idea of equivalent equations and equivalent inequalities. If we
perform an operation on an equation (or inequality) that produces another equation
(or inequality) with the same solution set, then the two equations (or inequalities) are
said to be equivalent. The basic idea in solving equations or inequalities is to per-
form operations that produce simpler equivalent equations or inequalities and to con-
tinue the process until we obtain an equation or inequality with an obvious solution.
Linear Equations
Linear equations are generally solved using the following equality properties.
THEOREM 1 Equality Properties

An equivalent equation will result if
1. The same quantity is added to or subtracted from each side of a given equation.
2. Each side of a given equation is multiplied by or divided by the same nonzero
quantity.
EXAMPLE 1 Solving a Linear Equation Solve and check:

8x - 31x - 42 = 31x - 42 + 6
SOLUTION 8x - 31x - 42 = 31x - 42 + 6 Use the distributive property.
8x - 3x + 12 = 3x - 12 + 6 Combine like terms.
5x + 12 = 3x - 6 Subtract 3x from both sides.
2x + 12 = - 6 Subtract 12 from both sides.
2x = - 18 Divide both sides by 2.
x = -9
532 APPENDIX A Linear Equations and Graphs
CHECK
8x - 31x - 42 = 31x - 42 + 6
81 ∙ 92 - 331 ∙ 92 - 44 ≟ 331 ∙ 92 - 44 + 6
- 72 - 31 - 132 ≟ 31 - 132 + 6
✓
- 33 = - 33
Matched Problem 1 Solve and check: 3x - 212x - 52 = 21x + 32 - 8

According to equality property 2, multiplying both sides of an equation by a nonzero
number always produces an equivalent equation. What is the smallest positive num-
ber that you could use to multiply both sides of the following equation to produce an
equivalent equation without fractions?
x + 1 x 1
- =
3 4 2
x + 2 x
EXAMPLE 2 Solving a Linear Equation Solve and check: - = 5
2 3
SOLUTION What operations can we perform on

x + 2 x
- = 5
2 3
to eliminate the denominators? If we can find a number that is exactly divisible by
each denominator, we can use the multiplication property of equality to clear the de-
nominators. The LCD (least common denominator) of the fractions, 6, is exactly what
we are looking for! Actually, any common denominator will do, but the LCD results
in a simpler equivalent equation. So we multiply both sides of the equation by 6:
Reminder
- b = 6#5
x + 2 x
Dashed boxes are used throughout the 6a
book to denote steps that are usually 2 3
1x + 22
6# - 6 # = 30
performed mentally. 3 2 x
2 3
1 1
31x + 22 - 2x = 30 Use the distributive property.

3x + 6 - 2x = 30 Combine like terms.
x + 6 = 30 Subtract 6 from both sides.
x = 24
x + 2 x
CHECK - = 5
2 3
24 + 2 24 ≟
- 5
2 3
13 - 8 ≟ 5
✓
5 =5
x + 1 x 1
Matched Problem 2 Solve and check: - =
3 4 2
In many applications of algebra, formulas or equations must be changed to alter-

native equivalent forms. The following example is typical.
EXAMPLE 3 Solving a Formula for a Particular Variable If you deposit a principal P in an ac-
count that earns simple interest at an annual rate r, then the amount A in the account
after t years is given by A = P + Prt. Solve for
(A) r in terms of A, P, and t
(B) P in terms of A, r, and t
SOLUTION (A) A = P + Prt Reverse equation.
P + Prt = A Subtract P from both sides.
Prt = A - P Divide both members by Pt.
A - P
r =
Pt
(B) A = P + Prt Reverse equation.
Factor out P (note the use
P + Prt = A
of the distributive property).
P11 + r t2 = A Divide by 11 + rt2.
A
P =
1 + rt
Matched Problem 3 If a cardboard box has length L, width W, and height H,
then its surface area is given by the formula S = 2LW + 2LH + 2WH. Solve the
formula for
(A) L in terms of S, W, and H (B) H in terms of S, L, and W
Linear Inequalities
Before we start solving linear inequalities, let us recall what we mean by 6 (less
than) and 7 (greater than). If a and b are real numbers, we write
a * b a is less than b
if there exists a positive number p such that a + p = b. Certainly, we would expect
that if a positive number was added to any real number, the sum would be larger than
the original. That is essentially what the definition states. If a 6 b, we may also write
b + a b is greater than a.
EXAMPLE 4 Inequalities Replace each question mark with either 6 or 7 .

(A) 3 ? 5 (B) - 6 ? - 2 (C) 0 ? - 10
SOLUTION
(A) 3 6 5 because 3 + 2 = 5.
(B) - 6 6 - 2 because - 6 + 4 = - 2.
(C) 0 7 - 10 because - 10 6 0 (because - 10 + 10 = 0).
Matched Problem 4 Replace each question mark with either 6 or 7.

(A) 2 ? 8 (B) - 20 ? 0 (C) - 3 ? - 30
a d b 0 c The inequality symbols have a very clear geometric interpretation on the real
Figure 1 a * b, c + d number line. If a 6 b, then a is to the left of b on the number line; if c 7 d, then c is
to the right of d on the number line (Fig. 1). Check this geometric property with the
inequalities in Example 4.

Replace ? with 6 or 7 in each of the following:
(A) - 1 ? 3 and 21 - 12 ? 2132
(B) - 1 ? 3 and - 21 - 12 ? - 2132
12 -8
(C) 12 ? - 8 and ?
4 4
12 -8
(D) 12 ? - 8 and ?
-4 -4
Based on these examples, describe the effect of multiplying both sides of an inequality
by a number.
The procedures used to solve linear inequalities in one variable are almost the
same as those used to solve linear equations in one variable, but with one important
exception, as noted in item 3 of Theorem 2.
THEOREM 2 Inequality Properties

An equivalent inequality will result, and the sense or direction will remain the
same, if each side of the original inequality
1. has the same real number added to or subtracted from it.
2. is multiplied or divided by the same positive number.
An equivalent inequality will result, and the sense or direction will reverse, if each
side of the original inequality
3. is multiplied or divided by the same negative number.
Note: Multiplication by 0 and division by 0 are not permitted.
Therefore, we can perform essentially the same operations on inequalities that we

perform on equations, with the exception that the sense of the inequality reverses
if we multiply or divide both sides by a negative number. Otherwise, the sense of
the inequality does not change. For example, if we start with the true statement
-3 7 -7
and multiply both sides by 2, we obtain
- 6 7 - 14
and the sense of the inequality stays the same. But if we multiply both sides of - 3 7 - 7
by - 2, the left side becomes 6 and the right side becomes 14, so we must write
6 6 14
to have a true statement. The sense of the inequality reverses.
If a 6 b, the double inequality a 6 x 6 b means that a * x and x * b; that
is, x is between a and b. Interval notation is also used to describe sets defined by
inequalities, as shown in Table 1.
The numbers a and b in Table 1 are called the endpoints of the interval. An in-
terval is closed if it contains all its endpoints and open if it does not contain any of
its endpoints. The intervals 3a, b4, 1- ∞, a4, and 3b, ∞2 are closed, and the intervals
1a, b2, 1- ∞, a2, and 1b, ∞ 2 are open. Note that the symbol ∞ (read infinity) is not a
number. When we write 3b, ∞2, we are simply referring to the interval that starts at
b and continues indefinitely to the right. We never refer to ∞ as an endpoint, and we
never write 3b, ∞ 4. The interval 1 - ∞, ∞ 2 is the entire real number line.
Table 1 Interval Notation

Interval Notation Inequality Notation Line Graph
3a, b4 a … x … b x
a b
3a, b2 a … x 6 b x
a b
1a, b4 a 6 x … b x
a b
1a, b2 a 6 x 6 b x
a b
1 - ∞ , a4 x … a x
a
1 - ∞ , a2 x 6 a x
a
3b, ∞ 2 x Ú b x
b
1b, ∞ 2 x 7 b x
b
Note that an endpoint of a line graph in Table 1 has a square bracket through it if
the endpoint is included in the interval; a parenthesis through an endpoint indicates
that it is not included.
CONCEPTUAL INSIGHT
The notation 12, 72 has two common mathematical interpretations: the ordered
pair with first coordinate 2 and second coordinate 7, and the open interval
consisting of all real numbers between 2 and 7. The choice of interpretation is
usually determined by the context in which the notation is used. The notation
12, - 72 could be interpreted as an ordered pair but not as an interval. In interval
notation, the left endpoint is always written first. So, 1 - 7, 22 is correct interval
notation, but 12, - 72 is not.
EXAMPLE 5 Interval and Inequality Notation, and Line Graphs

(A) Write 3 - 2, 32 as a double inequality and graph.
(B) Write x Ú - 5 in interval notation and graph.
SOLUTION
x
(A) 3 - 2, 32 is equivalent to - 2 … x 6 3. 22 3
(B) x Ú - 5 is equivalent to 3 - 5, ∞ 2. x
25
Matched Problem 5
(A) Write 1 - 7, 44 as a double inequality and graph.
(B) Write x 6 3 in interval notation and graph.

The solution to Example 5B shows the graph of the inequality x Ú - 5. What is the
graph of x 6 - 5? What is the corresponding interval? Describe the relationship be-
tween these sets.
EXAMPLE 6 Solving a Linear Inequality Solve and graph:

212x + 32 6 61x - 22 + 10
SOLUTION 212x + 32 6 61x - 22 + 10 Remove parentheses.

4x + 6 6 6x - 12 + 10 Combine like terms.
4x + 6 6 6x - 2 Subtract 6x from both sides.
- 2x + 6 6 - 2 Subtract 6 from both sides.
- 2x 6 - 8 Divide both sides by - 2 and
reverse the sense of the inequality.
x
x 7 4 or 14, ∞ 2 4
Notice that in the graph of x 7 4, we use a parenthesis through 4, since the

point 4 is not included in the graph.
Matched Problem 6 Solve and graph: 31x - 12 … 51x + 22 - 5
EXAMPLE 7 Solving a Double Inequality Solve and graph: - 3 6 2x + 3 … 9

SOLUTION We are looking for all numbers x such that 2x + 3 is between - 3 and 9,
including 9 but not - 3. We proceed as before except that we try to isolate x in the middle:
- 3 6 2x + 3 … 9
- 3 - 3 6 2x + 3 - 3 … 9 - 3
- 6 6 2x … 6
-6 2x 6
6 …
2 2 2
- 3 6 x … 3 or 1 - 3, 3] x
23 3
Matched Problem 7 Solve and graph: - 8 … 3x - 5 6 7
Note that a linear equation usually has exactly one solution, while a linear in-
equality usually has infinitely many solutions.
Applications
To realize the full potential of algebra, we must be able to translate real-world prob-
lems into mathematics. In short, we must be able to do word problems.
Here are some suggestions that will help you get started:
PROCEDURE For Solving Word Problems

1. Read the problem carefully and introduce a variable to represent an unknown
quantity in the problem. Often the question asked in a problem will indicate the
unknown quantity that should be represented by a variable.
2. Identify other quantities in the problem (known or unknown), and whenever pos-
sible, express unknown quantities in terms of the variable you introduced in Step 1.
3. Write a verbal statement using the conditions stated in the problem and then
write an equivalent mathematical statement (equation or inequality).
4. Solve the equation or inequality and answer the questions posed in the problem.
5. Check the solution(s) in the original problem.
EXAMPLE 8 Purchase Price Alex purchases a big screen TV, pays 7% state sales tax, and is
charged $65 for delivery. If Alex’s total cost is $1,668.93, what was the purchase
price of the TV?
SOLUTION
Step 1 Introduce a variable for the unknown quantity. After reading the problem,
we decide to let x represent the purchase price of the TV.
Step 2 Identify quantities in the problem.
Delivery charge: $65
Sales tax: 0.07x
Total cost: $1,668.93
Step 3 Write a verbal statement and an equation.
Price + Delivery Charge + Sales Tax = Total Cost

x + 65 + 0.07x = 1,668.93
Step 4 Solve the equation and answer the question.
x + 65 + 0.07x = 1,668.93 Combine like terms.
1.07x + 65 = 1,668.93 Subtract 65 from both sides.
1.07x = 1,603.93 Divide both sides by 1.07.
x = 1,499
The price of the TV is $1,499.
Step 5 Check the answer in the original problem.
Price = $1,499.00
Delivery charge = $ 65.00
Tax = 0.07 # 1,499 = $ 104.93
Total = $1,668.93
Matched Problem 8 Mary paid 8.5% sales tax and a $190 title and license fee
when she bought a new car for a total of $28,400. What is the purchase price of the car?
Any manufacturing company has costs, C, which include fixed costs such as
plant overhead, product design, setup, and promotion; and variable costs that depend
on the number of items produced. The revenue, R, is the amount of money received
from the sale of its product. The company breaks even if the revenue is equal to the
costs, that is, if R = C. Example 9 provides an introduction to cost, revenue, and
break-even analysis.
EXAMPLE 9 Break-Even Analysis A manufacturing company makes bike computers. Fixed

costs are $48,000, and variable costs are $12.40 per computer. If the computers are
sold at a price of $17.40 each, how many bike computers must be manufactured and
sold in order for the company to break even?
SOLUTION
Step 1 Let x = number of bike computers manufactured and sold.
Step 2 C = Fixed costs + Variable costs
= $48,000 + $12.40x
R = $17.40x
Step 3 The company breaks even if R = C; that is, if

$17.40x = $48,000 + $12.40x
Step 4 17.4x = 48,000 + 12.4x Subtract 12.4x from both sides.
5x = 48,000 Divide both sides by 5.
x = 9,600
The company must make and sell 9,600 bike computers to break even.
Step 5 Check:
Costs Revenue
48,000 + 12.419,6002 17.419,6002
= $167,040 = $167,040
Matched Problem 9 How many bike computers would a company have to

make and sell to break even if the fixed costs are $36,000, variable costs are $10.40
per computer, and the computers are sold to retailers for $15.20 each?
EXAMPLE 10 Consumer Price Index The Consumer Price Index (CPI) is a measure of the aver-
age change in prices over time from a designated reference period, which equals 100.
The index is based on prices of basic consumer goods and services. Table 2 lists the
Table 2 CPI 1 198291984 = 100 2 CPI for several years from 1960 to 2016. What net annual salary in 2016 would have
the same purchasing power as a net annual salary of $13,000 in 1960? Compute the
Year Index
answer to the nearest dollar. (Source: U.S. Bureau of Labor Statistics)
1960 29.6
1973 44.4 SOLUTION
1986 109.6 Step 1 Let x = the purchasing power of an annual salary in 2016.
1999 156.9
Step 2 Annual salary in 1960 = $13,000
2016 241.7
CPI in 1960 = 29.6
CPI in 2016 = 241.7
Step 3 The ratio of a salary in 2016 to a salary in 1960 is the same as the ratio of the
CPI in 2016 to the CPI in 1960.
x 241.7
= Multiply both sides by 13,000.
13,000 29.6
x = 13,000 #
241.7
Step 4
29.6
= $106,152 per year
Step 5 To check the answer, we confirm that the salary ratio agrees with the CPI ratio:
Salary Ratio CPI Ratio

106,152 241.7
= 8.166 = 8.166
13,000 29.6
Matched Problem 10 What net annual salary in 1973 would have had the same
purchasing power as a net annual salary of $100,000 in 2016? Compute the answer
to the nearest dollar.
Exercises A.1
A In Problems 1–6, solve for x. C In Problems 43 and 44, solve and graph.
1. 5x + 3 = x + 23 2. 7x - 6 = 5x - 24 43. - 3 … 4 - 7x 6 18
3. 914 - x2 = 21x + 72 4. 31x + 62 = 5 - 21x + 12 44. - 10 … 8 - 3u … - 6
x + 1 x 2x + 1 5x 45. If both a and b are positive numbers and b/a is greater than 1,
5. = + 5 6. - = 4
4 2 3 2 then is a - b positive or negative?
In Problems 7–12, write the interval as an inequality or double 46. If both a and b are negative numbers and b/a is greater than
inequality. 1, then is a - b positive or negative?
7. 34, 132 8. 1 - 3, 54
9. 1 - 2, 72 10. 3 - 6, - 14 Applications
11. 1 - ∞ , 44 12. 39, ∞ 2
47. Ticket sales. A rock concert brought in $432,500 on the sale
In Problems 13–18, write the solution set using interval notation. of 9,500 tickets. If the tickets sold for $35 and $55 each, how
13. - 8 6 x … 2 14. - 1 … x 6 5 many of each type of ticket were sold?
15. 2x 6 18 16. 3x Ú 12 48. Parking meter coins. An all-day parking meter takes only
dimes and quarters. If it contains 100 coins with a total value
17. 15 … - 3x 6 21 18. - 8 6 - 4x … 12 of $14.50, how many of each type of coin are in the meter?
B In Problems 19–32, find the solution set. 49. IRA. You have $500,000 in an IRA (Individual Retirement
Account) at the time you retire. You have the option of invest-
x 1 1 m 2 ing this money in two funds: Fund A pays 5.2% annually and
19. + = 20. - 4 = Fund B pays 7.7% annually. How should you divide your
4 2 8 3 3
money between Fund A and Fund B to produce an annual
y 3 x 5
21. 7 22. 6 interest income of $34,000?
-5 2 -4 6
23. 2u + 4 = 5u + 1 - 7u 50. IRA. Refer to Problem 49. How should you divide your
money between Fund A and Fund B to produce an annual
24. - 3y + 9 + y = 13 - 8y interest income of $30,000?
25. 10x + 251x - 32 = 275 51. Car prices. If the price change of cars parallels the change
26. - 314 - x2 = 5 - 1x + 12 in the CPI (see Table 2 in Example 10), what would a car sell
for (to the nearest dollar) in 2016 if a comparable model sold
27. 3 - y … 41y - 32 28. x - 2 Ú 21x - 52 for $10,000 in 1999?
x x 6 y y 1
29. - = 30. - = 52. Home values. If the price change in houses parallels the CPI
5 6 5 4 3 2 (see Table 2 in Example 10), what would a house valued at
m 3 m u 2 u $200,000 in 2016 be valued at (to the nearest dollar) in 1960?
31. - 3 6 - 32. - 6 + 2
5 5 2 2 3 3 53. Retail and wholesale prices. Retail prices in a department
store are obtained by marking up the wholesale price by 40%.
In Problems 33–36, solve and graph.
That is, the retail price is obtained by adding 40% of the
33. 2 … 3x - 7 6 14 34. - 4 … 5x + 6 6 21 wholesale price to the wholesale price.
9
35. - 4 … 5C + 32 … 68 36. - 1 … 23 t + 5 … 11 (A) What is the retail price of a suit if the wholesale price
is $300?
In Problems 37–42, solve for the indicated variable.
(B) What is the wholesale price of a pair of jeans if the retail
37. 3x - 4y = 12; for y price is $77?
38. y = - 23 x + 8; for x 54. Retail and sale prices. Sale prices in a department store
39. Ax + By = C; for y 1B ∙ 02 are obtained by marking down the retail price by 15%. That
is, the sale price is obtained by subtracting 15% of the retail
40. y = mx + b; for m price from the retail price.
41. F = 95 C + 32; for C (A) What is the sale price of a hat that has a retail price of $60?
42. C = 59 1F - 322; for F (B) What is the retail price of a dress that has a sale price
of $136?
55. Equipment rental. A golf course charges $52 for a round of (C) If the company wants to start making a profit at the same
golf using a set of their clubs, and $44 if you have your own production level as before the cost increase, how much
clubs. If you buy a set of clubs for $270, how many rounds should they sell the book for now?
must you play to recover the cost of the clubs? 62. Break-even analysis. The publisher in Problem 60 finds
56. Equipment rental. The local supermarket rents carpet clean- that rising prices for paper increase the variable costs to
ers for $20 a day. These cleaners use shampoo in a special $2.70 per book.
cartridge that sells for $16 and is available only from the su- (A) Discuss possible strategies the company might use to
permarket. A home carpet cleaner can be purchased for $300. deal with this increase in costs.
Shampoo for the home cleaner is readily available for $9 a
bottle. Past experience has shown that it takes two shampoo (B) If the company continues to sell the books for $15, how
cartridges to clean the 10-foot-by-12-foot carpet in your living many books must they sell now to make a profit?
room with the rented cleaner. Cleaning the same area with the
(C) If the company wants to start making a profit at the same
home cleaner will consume three bottles of shampoo. If you
production level as before the cost increase, how much
buy the home cleaner, how many times must you clean the
should they sell the book for now?
living-room carpet to make buying cheaper than renting?
63. Wildlife management. A naturalist estimated the total number
57. Sales commissions. One employee of a computer store is paid
of rainbow trout in a certain lake using the capture–mark–
a base salary of $2,000 a month plus an 8% commission on all
recapture technique. He netted, marked, and released 200
sales over $7,000 during the month. How much must the em-
rainbow trout. A week later, allowing for thorough mixing, he
ployee sell in one month to earn a total of $4,000 for the month?
again netted 200 trout, and found 8 marked ones among them.
58. Sales commissions. A second employee of the computer Assuming that the proportion of marked fish in the second sam-
store in Problem 57 is paid a base salary of $3,000 a month ple was the same as the proportion of all marked fish in the total
plus a 5% commission on all sales during the month. population, estimate the number of rainbow trout in the lake.
(A) How much must this employee sell in one month to earn 64. Temperature conversion. If the temperature for a 24-hour
a total of $4,000 for the month? period at an Antarctic station ranged between - 49°F and
14°F (that is, - 49 … F … 14), what was the range in de-
(B) Determine the sales level at which both employees
grees Celsius? [Note: F = 95 C + 32.]
receive the same monthly income.
65. Psychology. The IQ (intelligence quotient) is found by
(C) If employees can select either of these payment meth- dividing the mental age (MA), as indicated on standard tests,
ods, how would you advise an employee to make this by the chronological age (CA) and multiplying by 100. For
selection? example, if a child has a mental age of 12 and a chronologi-
59. Break-even analysis. A publisher for a promising new novel cal age of 8, the calculated IQ is 150. If a 9-year-old girl has
figures fixed costs (overhead, advances, promotion, copy an IQ of 140, compute her mental age.
editing, typesetting) at $55,000, and variable costs (printing, 66. Psychology. Refer to Problem 65. If the IQ of a group of
paper, binding, shipping) at $1.60 for each book produced. If 12-year-old children varies between 80 and 140, what is the
the book is sold to distributors for $11 each, how many must range of their mental ages?
be produced and sold for the publisher to break even?
60. Break-even analysis. The publisher of a new book figures Answers to Matched Problems
fixed costs at $92,000 and variable costs at $2.10 for each
book produced. If the book is sold to distributors for $15 1. x = 4 2. x = 2
each, how many must be sold for the publisher to break even? S - 2WH S - 2LW
3. (A) L = (B) H =
2W + 2H 2L + 2W
61. Break-even analysis. The publisher in Problem 59 finds that
rising prices for paper increase the variable costs to $2.10 per 4. (A) 6 (B) 6 (C) 7
book. 5. (A) - 7 6 x … 4; (B) 1 - ∞ , 32
x x
(A) Discuss possible strategies the company might use to 27 4 3
deal with this increase in costs.
6. x Ú - 4 or [- 4, ∞ 2 7. - 1 … x 6 4 or [- 1, 42
(B) If the company continues to sell the books for $11, how x x
many books must they sell now to make a profit? 24 21 4
8. $26,000 9. 7,500 bike computers

10. $18,370
SECTION A.2 Graphs and Lines 541
A.2 Graphs and Lines

■■ Cartesian Coordinate System In this section, we will consider one of the most basic geometric figures—a line.
■■ Graphs of Ax + By = C
When we use the term line in this book, we mean straight line. We will learn how to
recognize and graph a line and how to use information concerning a line to find its
■■ Slope of a Line equation. Examining the graph of any equation often results in additional insight into
■■ Equations of Lines: Special Forms the nature of the equation’s solutions.
■■ Applications
Cartesian Coordinate System
Recall that to form a Cartesian or rectangular coordinate system, we select two
Abscissa
y Ordinate
real number lines—one horizontal and one vertical—and let them cross through
II 10 I their origins as indicated in Figure 1. Up and to the right are the usual choices for
P 5 (a, b) the positive directions. These two number lines are called the horizontal axis and
b
Q 5 (25, 5) the vertical axis, or, together, the coordinate axes. The horizontal axis is usually
5 Coordinates referred to as the x axis and the vertical axis as the y axis, and each is labeled accord-
Origin ingly. The coordinate axes divide the plane into four parts called quadrants, which
x are numbered counterclockwise from I to IV (see Fig. 1).
0 5 a 10
210 25
Now we want to assign coordinates to each point in the plane. Given an arbitrary
Axis point P in the plane, pass horizontal and vertical lines through the point (Fig. 1). The
25
vertical line will intersect the horizontal axis at a point with coordinate a, and the
III IV horizontal line will intersect the vertical axis at a point with coordinate b. These two
210 numbers, written as the ordered pair 1a, b2, form the coordinates of the point P.
R 5 (10, 210)
The first coordinate, a, is called the abscissa of P; the second coordinate, b, is called
Figure 1 The Cartesian
the ordinate of P. The abscissa of Q in Figure 1 is - 5, and the ordinate of Q is 5.
(rectangular) coordinate system
The coordinates of a point can also be referenced in terms of the axis labels. The x
coordinate of R in Figure 1 is 10, and the y coordinate of R is - 10. The point with
coordinates 10, 02 is called the origin.
The procedure we have just described assigns to each point P in the plane a
unique pair of real numbers 1a, b2. Conversely, if we are given an ordered pair of
real numbers 1a, b2, then, reversing this procedure, we can determine a unique point
P in the plane. Thus,
There is a one-to-one correspondence between the points in a plane and
the elements in the set of all ordered pairs of real numbers.
This is often referred to as the fundamental theorem of analytic geometry.
Graphs of Ax ∙ By ∙ C
In Section A.1, we called an equation of the form ax + b = 0 1a ∙ 02 a linear
equation in one variable. Now we want to consider linear equations in two
variables:
DEFINITION Linear Equations in Two Variables

A linear equation in two variables is an equation that can be written in the
standard form
Ax + By = C
where A, B, and C are constants (A and B not both 0), and x and y are variables.
A solution of an equation in two variables is an ordered pair of real numbers that

satisfies the equation. For example, 14, 32 is a solution of 3x - 2y = 6. The
solution set of an equation in two variables is the set of all solutions of the equation.
The graph of an equation is the graph of its solution set.

(A) As noted earlier, 14, 32 is a solution of the equation
3x - 2y = 6
Find three more solutions of this equation. Plot these solutions in a Cartesian
coordinate system. What familiar geometric shape could be used to describe the
solution set of this equation?
(B) Repeat part (A) for the equation x = 2.
(C) Repeat part (A) for the equation y = - 3.
In Explore and Discuss 1, you may have recognized that the graph of each equation is
a (straight) line. Theorem 1 confirms this fact.
THEOREM 1 Graph of a Linear Equation in Two Variables

The graph of any equation of the form
Ax + By = C (A and B not both 0) (1)
is a line, and any line in a Cartesian coordinate system is the graph of an equation
of this form.
If A ∙ 0 and B ∙ 0, then equation (1) can be written as

A C
y = - x + = mx + b, m ∙ 0
B B
If A = 0 and B ∙ 0, then equation (1) can be written as
C
y =
B
and its graph is a horizontal line. If A ∙ 0 and B = 0, then equation (1) can be
written as
C
x =
Reminder A
If the x intercept is a and the y inter- and its graph is a vertical line. To graph equation (1), or any of its special cases, plot
cept is b, then the graph of the line any two points in the solution set and use a straightedge to draw the line through
passes through the points 1a, 02 and these two points. The points where the line crosses the axes are often the easiest to
10, b2. It is common practice to refer find. The y intercept is the y coordinate of the point where the graph crosses the
to both the numbers a and b and the y axis, and the x intercept is the x coordinate of the point where the graph crosses
points 1a, 02 and 10, b2 as the x and the x axis. To find the y intercept, let x = 0 and solve for y. To find the x intercept, let
y intercepts of the line. y = 0 and solve for x. It is a good idea to find a third point as a check point.
EXAMPLE 1 Using Intercepts to Graph a Line Graph: 3x - 4y = 12

SOLUTION Find and plot the y intercept, the x intercept, and a check point (Fig. 2).
x y y
0 -3 y intercept 10
4 0 x intercept
8 3 Check point Check point
5
(8, 3)
(4, 0)
x
210 25 5 10
(0, 23) x intercept
25
y intercept
210
Figure 2
Matched Problem 1 Graph: 4x - 3y = 12
The icon in the margin is used throughout this book to identify optional graphing
calculator activities that are intended to give you additional insight into the concepts
under discussion. You may have to consult the manual for your calculator for the
details necessary to carry out these activities.
EXAMPLE 2 Using a Graphing Calculator Graph 3x - 4y = 12 on a graphing calculator and

find the intercepts.
SOLUTION First, we solve 3x - 4y = 12 for y.
3x - 4y = 12 Add - 3x to both sides.

- 4y = - 3x + 12 Divide both sides by - 4.
- 3x + 12
y = Simplify.
-4
3
y = x - 3 (2)
4
Now we enter the right side of equation (2) in a calculator (Fig. 3A), enter values
for the window variables (Fig. 3B), and graph the line (Fig. 3C). (The numerals to
the left and right of the screen in Figure 3C are Xmin and Xmax, respectively. Simi-
larly, the numerals below and above the screen are Ymin and Ymax.)
210 10
25
(A) (B) (C)

Figure 3 Graphing a line on a graphing calculator
5 5
210 10 210 10
25 25
(A) (B)
Figure 4 Using trace and zero on a graphing calculator
Next we use two calculator commands to find the intercepts: trace (Fig. 4A) and
zero (Fig. 4B). The y intercept is - 3 (Fig. 4A), and the x intercept is 4 (Fig. 4B).
Matched Problem 2 Graph 4x - 3y = 12 on a graphing calculator and find

the intercepts.
EXAMPLE 3 Horizontal and Vertical Lines

(A) Graph x = - 4 and y = 6 simultaneously in the same rectangular coordinate
system.
(B) Write the equations of the vertical and horizontal lines that pass through the
point 17, - 52.
x 5 24 y
SOLUTION
10
(A) The line x = - 4 consists of all points with x coordinate - 4. To graph it, draw
the vertical line through 1 - 4, 02. The line y = 6 consists of all points with y
y56
5 coordinate 6. To graph it, draw the horizontal line through 10, 62. See Figure 5.
(B) Horizontal line through 17, - 52: y = - 5
x Vertical line through 17, - 52: x = 7
210 25 5 10
Matched Problem 3
25
(A) Graph x = 5 and y = - 3 simultaneously in the same rectangular coordinate
system.
210 (B) Write the equations of the vertical and horizontal lines that pass through the
Figure 5 point 1 - 8, 22.
Slope of a Line
If we take two points, 1x1, y1 2 and 1x2, y2 2, on a line, then the ratio of the change in
y to the change in x is called the slope of the line. In a sense, slope provides a mea-
sure of the “steepness” of a line relative to the x axis. The change in x is often called
the run, and the change in y is the rise.
y
P2 (x 2, y 2 ) DEFINITION Slope of a Line

If a line passes through two distinct points, 1x1, y1 2 and 1x2, y2 2 (see Fig. 6), then
y 2 2 y1 its slope is given by the formula
Rise
x
y2 - y1
P1(x1, y1) m = x1 ∙ x2
x2 - x1
(x 2, y1)
x 2 2 x1 vertical change 1rise2
Run
=
horizontal change 1run2
Figure 6
For a horizontal line, y does not change; its slope is 0. For a vertical line, x does not
change; x1 = x2 so its slope is not defined. In general, the slope of a line may be
positive, negative, 0, or not defined. Each case is illustrated geometrically in Table 1.
Table 1 Geometric Interpretation of Slope

Rising as x moves Falling as x moves Horizontal Vertical
Line from left to right from left to right
Slope Positive Negative 0 Not defined
Example
CONCEPTUAL INSIGHT
One property of real numbers discussed in Appendix B, Section B.1, is
-a -a a a
= - = - = , b ∙ 0
-b b -b b
This property implies that the slope of the line through A and B is equal to the slope
b of the line through B and A. For example, if A = 14, 32 and B = 11, 22, then
a B = 11, 22 A = 14, 32
a a9 A = 14, 32 B = 11, 22
m 5 2 5 22
b b9
b9
2 - 3 -1 1 3 - 2
m = = = =
1 - 4 -3 3 4 - 1
a9 A property of similar triangles (see the references at the back of the book) ensures
Figure 7
that the slope of a line is the same for any pair of distinct points on the line (Fig. 7).
EXAMPLE 4 Finding Slopes Sketch a line through each pair of points, and find the slope of
each line.
(A) 1 - 3, - 22, 13, 42 (B) 1 - 1, 32, 12, - 32
(C) 1 - 2, - 32, 13, - 32 (D) 1 - 2, 42, 1 - 2, - 22
SOLUTION
(A) y (B) y
5 5
x x
25 5 25 5
25 25
4 - 1 - 22 6 -3 - 3 -6
m = = = 1 m = = = -2
3 - 1 - 32 6 2 - 1 - 12 3
(C) y (D) y
5 5
x x
25 5 25 5
25 25
- 3 - 1 - 32 0 -2 - 4 -6
m = = = 0 m = =
3 - 1 - 22 5 - 2 - 1 - 22 0
Slope is not defined.
Matched Problem 4 Find the slope of the line through each pair of points.
(A) 1 - 2, 42, 13, 42 (B) 1 - 2, 42, 10, - 42
(C) 1 - 1, 52, 1 - 1, - 22 (D) 1 - 1, - 22, 12, 12
Equations of Lines: Special Forms

Let us start by investigating why y = mx + b is called the slope-intercept form for
a line.

(A) Graph y = x + b for b = - 5, - 3, 0, 3, and 5 simultaneously in the same coor-
dinate system. Verbally describe the geometric significance of b.
(B) Graph y = mx - 1 for m = - 2, - 1, 0, 1, and 2 simultaneously in the same
coordinate system. Verbally describe the geometric significance of m.
(C) Using a graphing calculator, explore the graph of y = mx + b for different
values of m and b.
As you may have deduced from Explore and Discuss 2, constants m and b in
y = mx + b have the following geometric interpretations.
If we let x = 0, then y = b. So the graph of y = mx + b crosses the y axis at
10, b2. The constant b is the y intercept. For example, the y intercept of the graph of
y = - 4x - 1 is - 1.
To determine the geometric significance of m, we proceed as follows: If
y = mx + b, then by setting x = 0 and x = 1, we conclude that 10, b2 and
11, m + b2 lie on its graph (Fig. 8). The slope of this line is given by:
y2 - y1 1m + b2 - b
Slope = = = m
x2 - x1 1 - 0
So m is the slope of the line given by y = mx + b.
(1, m 1 b)
(0, b)
x
Figure 8
DEFINITION Slope-Intercept Form

The equation
y = mx + b m = slope, b = y intercept (3)
is called the slope-intercept form of an equation of a line.
EXAMPLE 5 Using the Slope-Intercept Form

(A) Find the slope and y intercept, and graph y = - 23 x - 3.
(B) Write the equation of the line with slope 23 and y intercept - 2.
SOLUTION
(A) Slope = m = - 23; y intercept = b = - 3.
To graph the line, first plot the y intercept 10, - 32. Then, since the slope is - 23,
locate a second point on the line by moving 3 units in the x direction (run) and - 2
units in the y direction (rise). Draw the line through these two points (see Fig. 9).
y
x
25 5
3
y intercept 22
25
Figure 9
2 2
(B) m = 3 and b = - 2; so, y = 3 x - 2
Matched Problem 5 Write the equation of the line with slope 12 and y intercept
- 1. Graph.
y Suppose that a line has slope m and passes through a fixed point 1x1, y1 2. If the
point 1x, y2 is any other point on the line (Fig. 10), then
(x, y)
y - y1
= m
x - x1
That is,
x
y - y1 = m1x - x1 2 (4)
(x1, y1) (x, y1)
We now observe that 1x1, y1 2 also satisfies equation (4) and conclude that equa-
Figure 10 tion (4) is an equation of a line with slope m that passes through 1x1, y1 2.
DEFINITION Point-Slope Form

An equation of a line with slope m that passes through 1x1, y1 2 is
y - y1 = m1x - x1 2 (4)
which is called the point-slope form of an equation of a line.
The point-slope form is extremely useful, since it enables us to find an equation for a
line if we know its slope and the coordinates of a point on the line or if we know the
coordinates of two points on the line.
EXAMPLE 6 Using the Point-Slope Form

(A) Find an equation for the line that has slope 12 and passes through 1 - 4, 32. Write
the final answer in the form Ax + By = C.
(B) Find an equation for the line that passes through the points 1 - 3, 22 and
1 - 4, 52. Write the resulting equation in the form y = mx + b.
SOLUTION
1
(A) Use y - y1 = m1x - x1 2. Let m = 2 and 1x1, y1 2 = 1 - 4, 32. Then
y - 3 = 12[x - 1 - 42]
y - 3 = 12 1x + 42 Multiply both sides by 2.
2y - 6 = x + 4
- x + 2y = 10 or x - 2y = - 10
(B) First, find the slope of the line by using the slope formula:
y2 - y1 5 - 2 3
m = = = = -3
x2 - x1 - 4 - 1 - 32 -1
Now use y - y1 = m1x - x1 2 with m = - 3 and 1x1, y1 2 = 1 - 3, 22:
y - 2 = - 33x - 1 - 324
y - 2 = - 31x + 32
y - 2 = - 3x - 9
y = - 3x - 7
Matched Problem 6
(A) Find an equation for the line that has slope 23 and passes through 16, - 22.
Write the resulting equation in the form Ax + By = C, A 7 0.
(B) Find an equation for the line that passes through 12, - 32 and 14, 32. Write
the resulting equation in the form y = mx + b.
The various forms of the equation of a line that we have discussed are summa-
rized in Table 2 for quick reference.
Table 2 Equations of a Line

Standard form Ax + By = C A and B not both 0
Slope-intercept form y = mx + b Slope: m; y intercept: b
Point-slope form y - y1 = m1x - x1 2 Slope: m; point: 1x1, y1 2
Horizontal line y = b Slope: 0
Vertical line x = a Slope: undefined
Applications
We will now see how equations of lines occur in certain applications.
EXAMPLE 7 Cost Equation The management of a company that manufactures skateboards has
fixed costs (costs at 0 output) of $300 per day and total costs of $4,300 per day at an
output of 100 skateboards per day. Assume that cost C is linearly related to output x.
(A) Find the slope of the line joining the points associated with outputs of 0 and
100, that is, the line passing through 10, 3002 and 1100, 4,3002.
(B) Find an equation of the line relating output to cost. Write the final answer in the
form C = mx + b.
(C) Graph the cost equation from part (B) for 0 … x … 200.
SOLUTION
y2 - y1
(A) m =
x2 - x1
4,300 - 300
=
100 - 0
4,000
= = 40
100
C
(B) We must find an equation of the line that passes through 10, 3002 with slope
10,000 40. We use the slope-intercept form:
C 5 40x 1 300
Cost per Day (dollars)
8,000 C = mx + b
6,000 C = 40x + 300
4,000 (C) To graph C = 40x + 300 for 0 … x … 200, we first calculate C12002:
2,000 C12002 = 4012002 + 300
x = 8,300
100 200
Output per Day Plot the points 1200, 8,3002 and 10, 3002 and draw the line segment joining
Figure 11 them (see Fig. 11).
In Example 7, the fixed cost of $300 per day covers plant cost, insurance, and so
on. This cost is incurred whether or not there is any production. The variable cost is
40x, which depends on the day’s output. Since increasing production from x to x + 1
will increase the cost by $40 (from 40x + 300 to 40x + 340), the slope 40 can be
interpreted as the rate of change of the cost function with respect to production x.
Matched Problem 7 Answer parts (A) and (B) in Example 7 for fixed costs
of $250 per day and total costs of $3,450 per day at an output of 80 skateboards
per day.
In a free competitive market, the price of a product is determined by the relation-

ship between supply and demand. If there is a surplus—that is, the supply is greater
than the demand—the price tends to come down. If there is a shortage—that is, the
demand is greater than the supply—the price tends to go up. The price tends to move
toward an equilibrium price at which the supply and demand are equal. Example 8
introduces the basic concepts.
EXAMPLE 8 Supply and Demand At a price of $9.00 per box of oranges, the supply is 320,000
boxes and the demand is 200,000 boxes. At a price of $8.50 per box, the supply is
270,000 boxes and the demand is 300,000 boxes.
(A) Find a price–supply equation of the form p = mx + b, where p is the price in
dollars and x is the corresponding supply in thousands of boxes.
(B) Find a price–demand equation of the form p = mx + b, where p is the price in
dollars and x is the corresponding demand in thousands of boxes.
(C) Graph the price–supply and price–demand equations in the same coordinate
system and find their point of intersection.
SOLUTION
(A) To find a price–supply equation of the form p = mx + b, we must find two
points of the form 1x, p2 that are on the supply line. From the given supply data,
1320, 92 and 1270, 8.52 are two such points. First, find the slope of the line:
9 - 8.5 0.5
m = = = 0.01
320 - 270 50
Now use the point-slope form to find the equation of the line:
p - p1 = m1x - x1 2 1x1, p1 2 = 1320, 92
p - 9 = 0.011x - 3202
p - 9 = 0.01x - 3.2
(B) From the given demand data, 1200, 92 and 1300, 8.52 are two points on the
demand line.
8.5 - 9 - 0.5
m = = = - 0.005
300 - 200 100
p - p1 = m1x - x1 2 Substitute x1 = 200, p1 = 9.
p - 9 = - 0.0051x - 2002
p - 9 = - 0.005x + 1
(C) From part (A), we plot the points 1320, 92 and 1270, 8.52 and then draw the
line through them. We do the same with the points 1200, 92 and 1300, 8.52 from
part (B) (Fig. 12). (Note that we restricted the axes to intervals that contain these
data points.) To find the intersection point of the two lines, we equate the right-
hand sides of the price–supply and price–demand equations and solve for x:
Price9supply Price9demand
0.01x + 5.8 = - 0.005x + 10
0.015x = 4.2
x = 280
p Price–supply equation
p 5 0.01x 1 5.8
9.00 (200, 9)
(320, 9)
Equilibrium point
Equilibrium (280, 8.6)
8.60
price ($) (300, 8.5)
8.50 (270, 8.5)
Price–demand equation
p 5 20.005x 1 10
8.00
x
0 200 240 280 320
Equilibrium
quantity
Figure 12 Graphs of price–supply and price–demand equations
Now use the price–supply equation to find p when x = 280:

p = 0.01x + 5.8
p = 0.0112802 + 5.8 = 8.6
As a check, we use the price–demand equation to find p when x = 280:
p = - 0.005x + 10
10
p = - 0.00512802 + 10 = 8.6
The lines intersect at 1280, 8.62. The intersection point of the price–supply and
price–demand equations is called the equilibrium point, and its coordinates are
the equilibrium quantity (280) and the equilibrium price ($8.60). These terms
180 360 are illustrated in Figure 12. The intersection point can also be found by using the
7
intersect command on a graphing calculator (Fig. 13). To summarize, the price of
Figure 13 Finding an intersection a box of oranges tends toward the equilibrium price of $8.60, at which the supply
point and demand are both equal to 280,000 boxes.
Matched Problem 8 At a price of $12.59 per box of grapefruit, the supply is

595,000 boxes and the demand is 650,000 boxes. At a price of $13.19 per box, the
supply is 695,000 boxes and the demand is 590,000 boxes. Assume that the rela-
tionship between price and supply is linear and that the relationship between price
and demand is linear.
(A) Find a price–supply equation of the form p = mx + b.
(B) Find a price–demand equation of the form p = mx + b.
(C) Find the equilibrium point.
Exercises A.2
A Problems 1–4 refer to graphs (A)–(D). 3. Identify the graph(s) of any lines with slope zero.
y y
4. Identify the graph(s) of any lines with undefined slope.
5 5
In Problems 5–8, sketch a graph of each equation in a rectangular
coordinate system.
x
x x 5. y = 2x - 3 6. y = + 1
25 5 25 5 2
7. 2x + 3y = 12 8. 8x - 3y = 24
25 25 In Problems 9–14, find the slope and y intercept of the graph of

each equation.
(A) (B)
9. y = 5x - 7 10. y = 3x + 2
y y 5 10
11. y = - x - 9 12. y = - x + 4
5 5 2 3
x 2 x 1
13. y = + 14. y = -
4 3 5 2
x x
25 5 25 5
In Problems 15–20, find the slope and x intercept of the graph of
each equation.
25 25 15. y = 2x + 10 16. y = - 4x + 12
(C) (D) 17. 8x - y = 40 18. 3x + y = 6
19. - 6x + 7y = 42 20. 9x + 2y = 4
1. Identify the graph(s) of lines with a negative slope.
2. Identify the graph(s) of lines with a positive slope.
In Problems 21–24, write an equation of the line with the indi- 43. Graph y = 25x + 200, x Ú 0.
cated slope and y intercept.
44. Graph y = 40x + 160, x Ú 0.
21. Slope = 2 22. Slope = 1 45. (A) Graph y = 1.2x - 4.2 in a rectangular coordinate system.
y intercept = 1 y intercept = 5
(B) Find the x and y intercepts algebraically to one decimal
1 6 place.
23. Slope = - 24. Slope =
3 7 (C) Graph y = 1.2x - 4.2 in a graphing calculator.
9
y intercept = 6 y intercept = - (D) Find the x and y intercepts to one decimal place using
2
trace and the zero command.
B In Problems 25–28, use the graph of each line to find the x inter- 46. (A) Graph y = - 0.8x + 5.2 in a rectangular coordinate
cept, y intercept, and slope. Write the slope-intercept form of the system.
equation of the line.
(B) Find the x and y intercepts algebraically to one decimal
25. y 26. y place.
5 5 (C) Graph y = - 0.8x + 5.2 in a graphing calculator.
(D) Find the x and y intercepts to one decimal place using
trace and the zero command.
x x
25 5 25 5 In Problems 47–50, write the equations of the vertical and hori-
zontal lines through each point.
47. 14, - 32 48. 1 - 5, 62
25 25
49. 1 - 1.5, - 3.52 50. 12.6, 3.82
27. y 28. y
C In Problems 51–58, write the slope-intercept form of the equation of
5 5 the line with the indicated slope that goes through the given point.
51. m = 5; 13, 02 52. m = 4; 10, 62
x x
53. m = - 2; 1 - 1, 92 54. m = - 10; 12, - 52
25 5 25 5
1 2
55. m = ; 1 - 4, - 82 56. m = ; 17, 12
3 7
25 25 57. m = - 3.2; 15.8, 12.32 58. m = 0.9; 12.3, 6.72
In Problems 29–34, sketch a graph of each equation or pair of In Problems 59–66,

equations in a rectangular coordinate system. (A) Find the slope of the line that passes through the given
29. y = - 23 x - 2 30. y = - 32 x + 1 points.
31. 3x - 2y = 10 32. 5x - 6y = 15 (B) Find the standard form of the equation of the line.
33. x = 3; y = - 2 34. x = - 3; y = 2 (C) Find the slope-intercept form of the equation of the line.
In Problems 35–40, find the slope of the graph of each equation. 59. 12, 52 and 15, 72 60. 11, 22 and 13, 52
35. 4x + y = 3 36. 5x - y = - 2 61. 1 - 2, - 12 and 12, - 62 62. 12, 32 and 1 - 3, 72

63. 15, 32 and 15, - 32 64. 11, 42 and 10, 42
37. 3x + 5y = 15 38. 2x - 3y = 18
65. 1 - 2, 52 and 13, 52 66. 12, 02 and 12, - 32
39. - 4x + 2y = 9 40. - x + 8y = 4
67. Discuss the relationship among the graphs of the lines with
41. Given Ax + By = 12, graph each of the following three
equation y = mx + 2, where m is any real number.
cases in the same coordinate system.
(A) A = 2 and B = 0 68. Discuss the relationship among the graphs of the lines with
equation y = - 0.5x + b, where b is any real number.
(B) A = 0 and B = 3
(C) A = 3 and B = 4
42. Given Ax + By = 24, graph each of the following three Applications
cases in the same coordinate system.
(A) A = 6 and B = 0 69. Cost analysis. A donut shop has a fixed cost of $124 per
day and a variable cost of $0.12 per donut. Find the total
(B) A = 0 and B = 8
daily cost of producing x donuts. How many donuts can be
(C) A = 2 and B = 3 produced for a total daily cost of $250?
70. Cost analysis. A small company manufactures picnic tables. (B) What is the depreciated value of the boat after 10 years?
The weekly fixed cost is $1,200 and the variable cost is $45 (C) When will the depreciated value fall below $100,000?
per table. Find the total weekly cost of producing x picnic
tables. How many picnic tables can be produced for a total (D) Graph V for 0 … t … 30 and illustrate the answers from
weekly cost of $4,800? (B) and (C) on the graph.
71. Cost analysis. A plant can manufacture 80 golf clubs per 77. Boiling point. The temperature at which water starts to
day for a total daily cost of $7,647 and 100 golf clubs per day boil is called its boiling point and is linearly related to the
for a total daily cost of $9,147. altitude. Water boils at 212°F at sea level and at 193.6°F at an
altitude of 10,000 feet. (Source: biggreenegg.com)
(A) Assuming that daily cost and production are linearly re-
lated, find the total daily cost of producing x golf clubs. (A) Find a relationship of the form T = mx + b where T is
degrees Fahrenheit and x is altitude in thousands of feet.
(B) Graph the total daily cost for 0 … x … 200.
(B) Find the boiling point at an altitude of 3,500 feet.
(C) Interpret the slope and y intercept of this cost equation.
(C) Find the altitude if the boiling point is 200°F.
72. Cost analysis. A plant can manufacture 50 tennis rackets per
day for a total daily cost of $3,855 and 60 tennis rackets per (D) Graph T and illustrate the answers to (B) and (C) on
day for a total daily cost of $4,245. the graph.
(A) Assuming that daily cost and production are linearly 78. Boiling point. The temperature at which water starts to boil
related, find the total daily cost of producing x tennis is also linearly related to barometric pressure. Water boils at
rackets. 212°F at a pressure of 29.9 inHg (inches of mercury) and at
191°F at a pressure of 28.4 inHg. (Source: biggreenegg.com)
(B) Graph the total daily cost for 0 … x … 100.
(A) Find a relationship of the form T = mx + b, where
(C) Interpret the slope and y intercept of this cost equation.
T is degrees Fahrenheit and x is pressure in inches of
73. Business—Markup policy. A drugstore sells a drug costing mercury.
$85 for $112 and a drug costing $175 for $238.
(B) Find the boiling point at a pressure of 31 inHg.
(A) If the markup policy of the drugstore is assumed to be
(C) Find the pressure if the boiling point is 199°F.
linear, write an equation that expresses retail price R in
terms of cost C (wholesale price). (D) Graph T and illustrate the answers to (B) and (C) on the
graph.
(B) What does a store pay (to the nearest dollar) for a drug
that retails for $185? 79. Flight conditions. In stable air, the air temperature drops
about 3.6°F for each 1,000-foot rise in altitude. (Source:
74. Business—Markup policy. A clothing store sells a shirt
Federal Aviation Administration)
costing $20 for $33 and a jacket costing $60 for $93.
(A) If the temperature at sea level is 70°F, write a linear
(A) If the markup policy of the store is assumed to be linear,
equation that expresses temperature T in terms of alti-
write an equation that expresses retail price R in terms of
tude A in thousands of feet.
cost C (wholesale price).
(B) At what altitude is the temperature 34°F?
(B) What does a store pay for a suit that retails for $240?
80. Flight navigation. The airspeed indicator on some aircraft
75. Business—Depreciation. A farmer buys a new tractor for
is affected by the changes in atmospheric pressure at dif-
$157,000 and assumes that it will have a trade-in value of
ferent altitudes. A pilot can estimate the true airspeed by
$82,000 after 10 years. The farmer uses a constant rate of
observing the indicated airspeed and adding to it about
depreciation (commonly called straight-line depreciation—
1.6% for every 1,000 feet of altitude. (Source: Megginson
one of several methods permitted by the IRS) to determine
Technologies Ltd.)
the annual value of the tractor.
(A) A pilot maintains a constant reading of 200 miles per
(A) Find a linear model for the depreciated value V of the
hour on the airspeed indicator as the aircraft climbs from
tractor t years after it was purchased.
sea level to an altitude of 10,000 feet. Write a linear
(B) What is the depreciated value of the tractor after 6 years? equation that expresses true airspeed T (in miles per
hour) in terms of altitude A (in thousands of feet).
(C) When will the depreciated value fall below $70,000?
(B) What would be the true airspeed of the aircraft at
(D) Graph V for 0 … t … 20 and illustrate the answers from
6,500 feet?
parts (B) and (C) on the graph.
81. Demographics. The average number of persons per house-
76. Business—Depreciation. A charter fishing company buys
hold in the United States has been shrinking steadily for as
a new boat for $224,000 and assumes that it will have a trade-
long as statistics have been kept and is approximately linear
in value of $115,200 after 16 years.
with respect to time. In 1980 there were about 2.76 persons
(A) Find a linear model for the depreciated value V of the per household, and in 2015 about 2.54. (Source: U.S. Census
boat t years after it was purchased. Bureau)
(A) If N represents the average number of persons per house- (B) Find a price–demand equation of the form p = mx + b.
hold and t represents the number of years since 1980,
write a linear equation that expresses N in terms of t.
(D) Graph the price–supply equation, price–demand equa-
(B) Use this equation to estimate household size in the
tion, and equilibrium point in the same coordinate
year 2030.
system.
82. Demographics. The median household income divides the
87. Physics. Hooke’s law states that the relationship between the
households into two groups: the half whose income is less
stretch s of a spring and the weight w causing the stretch is
than or equal to the median, and the half whose income is
linear. For a particular spring, a 5-pound weight causes a stretch
greater than the median. The median household income in
of 2 inches, while with no weight, the stretch of the spring is 0.
the United States grew from about $30,000 in 1990 to about
$55,775 in 2015. (Source: U.S. Census Bureau) (A) Find a linear equation that expresses s in terms of w.
(A) If I represents the median household income and t rep- (B) What is the stretch for a weight of 20 pounds?
resents the number of years since 1990, write a linear
(C) What weight will cause a stretch of 3.6 inches?
equation that expresses I in terms of t.
88. Physics. The distance d between a fixed spring and the floor
(B) Use this equation to estimate median household income
is a linear function of the weight w attached to the bottom
in the year 2030.
of the spring. The bottom of the spring is 18 inches from the
83. Cigarette smoking. The percentage of female cigarette floor when the weight is 3 pounds, and 10 inches from the
smokers in the United States declined from 21.0% in 2000 to floor when the weight is 5 pounds.
13.6% in 2015. (Source: Centers for Disease Control)
(A) Find a linear equation that expresses d in terms of w.
(A) Find a linear equation relating percentage of female
(B) Find the distance from the bottom of the spring to the
smokers (f) to years since 2000 (t).
floor if no weight is attached.
(B) Use this equation to predict the year in which the per-
(C) Find the smallest weight that will make the bottom of
centage of female smokers falls below 7%.
the spring touch the floor. (Ignore the height of the
84. Cigarette smoking. The percentage of male cigarette weight.)
smokers in the United States declined from 25.7% in 2000 to
16.7% in 2015. (Source: Centers for Disease Control) Answers to Matched Problems
(A) Find a linear equation relating percentage of male smok- 1. y
ers (m) to years since 2000 (t). 5
(B) Use this equation to predict the year in which the per- 25 5
x
centage of male smokers falls below 7%.

25
85. Supply and demand. At a price of $9.00 per bushel,
the supply of soybeans is 3,600 million bushels and the 2. y intercept = - 4, x intercept = 3
demand is 4,000 million bushels. At a price of $9.50 per
bushel, the supply is 4,100 million bushels and the demand is
5
3,500 million bushels.
(A) Find a price–supply equation of the form p = mx + b.
(B) Find a price–demand equation of the form p = mx + b. 25 5

(D) Graph the price–supply equation, price–demand equa- 25
tion, and equilibrium point in the same coordinate
system.
5
86. Supply and demand. At a price of $3.20 per bushel, the
supply of corn is 9,800 million bushels and the demand is
9,200 million bushels. At a price of $2.95 per bushel, the sup-
25 5
ply is 9,300 million bushels and the demand is 9,700 million
bushels.
(A) Find a price–supply equation of the form p = mx + b. 25
SECTION A.3 Linear Regression 555
3. (A) y 5. y = 12x - 1
x55
10 y
5 5
x
210 25 5 10
25 y 5 23 x
25 5
210
(B) Horizontal line: y = 2; 25

vertical line: x = - 8
6. (A) 2x - 3y = 18 (B) y = 3x - 9
4. (A) 0 (B) - 4
7. (A) m = 40 (B) C = 40x + 250
(C) Not defined (D) 1
8. (A) p = 0.006x + 9.02 (B) p = - 0.01x + 19.09
(C) 1629, 12.802
A.3 Linear Regression

■■ Slope as a Rate of Change Mathematical modeling is the process of using mathematics to solve real-world prob-
■■ Linear Regression
lems. This process can be broken down into three steps (Fig. 1):
Step 1 Construct the mathematical model (that is, a mathematics problem that,
when solved, will provide information about the real-world problem).
Step 2 Solve the mathematical model.
Step 3 Interpret the solution to the mathematical model in terms of the original real-
world problem.
Real-world
problem
1.
ret
Co
erp
nst
Int
ruc
3.
Mathematical Mathematical
2. Solve
solution model
Figure 1
In more complex problems, this cycle may have to be repeated several times to ob-
tain the required information about the real-world problem. In this section, we will
discuss one of the simplest mathematical models, a linear equation. With the aid of
a graphing calculator or computer, we also will learn how to analyze a linear model
based on real-world data.
Slope as a Rate of Change

If x and y are related by the equation y = mx + b, where m and b are constants with
m ∙ 0, then x and y are linearly related. If 1x1, y1 2 and 1x2, y2 2 are two distinct
points on this line, then the slope of the line is
y2 - y1 Change in y
m = = (1)
x2 - x1 Change in x
In applications, ratio (1) is called the rate of change of y with respect to x. Since
the slope of a line is unique, the rate of change of two linearly related variables is
constant. Here are some examples of familiar rates of change: miles per hour, revolu-
tions per minute, price per pound, passengers per plane, and so on. If the relationship
between x and y is not linear, ratio (1) is called the average rate of change of y with
respect to x.
EXAMPLE 1 Estimating Body Surface Area Appropriate doses of medicine for both animals
and humans are often based on body surface area (BSA). Since weight is much
easier to determine than BSA, veterinarians use the weight of an animal to estimate
BSA. The following linear equation expresses BSA for canines in terms of weight:
a = 16.12w + 375.6
where a is BSA in square inches and w is weight in pounds. (Source: Veterinary
Oncology Consultants, PTY LTD)
(A) Interpret the slope of the BSA equation.
(B) What is the effect of a one-pound increase in weight?
SOLUTION
(A) The rate-of-change of BSA with respect to weight is 16.12 square inches per
pound.
(B) Since slope is the ratio of rise to run, increasing w by 1 pound (run) increases a
by 16.12 square inches (rise).
Matched Problem 1 The equation a = 28.55w + 118.7 expresses BSA for

felines in terms of weight, where a is BSA in square inches and w is weight in
pounds.
(A) Interpret the slope of the BSA equation.
(B) What is the effect of a one-pound increase in weight?

As illustrated in Example 1A, the slope m of a line with equation y = mx + b has
two interpretations:
1. m is the rate of change of y with respect to x.
2. Increasing x by one unit will change y by m units.
How are these two interpretations related?
Parachutes are used to deliver cargo to areas that cannot be reached by other
means. The rate of descent of the cargo is the rate of change of altitude with respect
to time. The absolute value of the rate of descent is called the speed of the cargo. At
low altitudes, the altitude of the cargo and the time in the air are linearly related. The
appropriate rate of descent varies widely with the item. Bulk food (rice, flour, beans,
etc.) and clothing can tolerate nearly any rate of descent under 40 ft>sec. Machinery
and electronics (pumps, generators, radios, etc.) should generally be dropped at
15 ft>sec or less. Butler Tactical Parachute Systems in Roanoke, Virginia, manufac-
tures a variety of canopies for dropping cargo. The following example uses informa-
tion taken from the company’s brochures.
EXAMPLE 2 Finding the Rate of Descent A 100-pound cargo of delicate electronic equip-
ment is dropped from an altitude of 2,880 feet and lands 200 seconds later. (Source:
Butler Tactical Parachute Systems)
(A) Find a linear model relating altitude a (in feet) and time in the air t (in seconds).
(B) How fast is the cargo moving when it lands?
SOLUTION
(A) If a = mt + b is the linear equation relating altitude a and time in air t, then
the graph of this equation must pass through the following points:
1t1, a1 2 = 10, 2,8802 Cargo is dropped from plane.
1t2, a2 2 = 1200, 02 Cargo lands.
The slope of this line is
a2 - a1 0 - 2,880
m = = = - 14.4
t2 - t1 200 - 0
and the equation of this line is
a - 0 = - 14.41t - 2002
a = - 14.4t + 2,880
(B) The rate of descent is the slope m = - 14.4, so the speed of the cargo at land-
ing is 0 - 14.4 0 = 14.4 ft>sec.
Matched Problem 2 A 400-pound load of grain is dropped from an altitude of

2,880 feet and lands 80 seconds later.
(A) Find a linear model relating altitude a (in feet) and time in the air t (in seconds).
(B) How fast is the cargo moving when it lands?
Linear Regression
In real-world applications, we often encounter numerical data in the form of a table.
Regression analysis is a process for finding a function that provides a useful model
for a set of data points. Graphs of equations are often called curves, and regression
analysis is also referred to as curve fitting. In the next example, we use a linear
model obtained by using linear regression on a graphing calculator.
EXAMPLE 3 Diamond Prices Prices for round-shaped diamonds taken from an online trader
are given in Table 1.
(A) A linear model for the data in Table 1 is given by
Table 1 Round-Shaped p = 6,140c - 480 (2)
Diamond Prices
where p is the price of a diamond weighing c carats. (We will discuss the source
Weight (carats) Price of models like this later in this section.) Plot the points in Table 1 on a Cartesian
0.5 $2,790 coordinate system, producing a scatter plot, and graph the model on the same
0.6 $3,191 axes.
0.7 $3,694
(B) Interpret the slope of the model in (2).
0.8 $4,154
0.9 $5,018 (C) Use the model to estimate the cost of a 0.85-carat diamond and the cost of
1.0 $5,898 a 1.2-carat diamond. Round answers to the nearest dollar.
Source: www.tradeshop.com (D) Use the model to estimate the weight of a diamond (to two decimal places) that
sells for $4,000.
SOLUTION
(A) A scatter plot is simply a graph of the points in Table 1 (Fig. 2A). To add the
graph of the model to the scatter plot, we find any two points that satisfy equa-
tion (2) 3we choose 10.4, 1,9762 and 11.1, 6,27424. Plotting these points and
drawing a line through them gives us Figure 2B.
p p
8000 8000
7000 7000
6000 6000
Price (dollars)
Price (dollars)
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
c c
0 0.2 0.4 0.6 0.8 1.0 1.2 0 0.2 0.4 0.6 0.8 1.0 1.2
Carats Carats
(A) Scatter plot (B) Linear model
Figure 2
(B) The rate of change of the price of a diamond with respect to its weight is 6,140.
Increasing the weight by one carat will increase the price by about $6,140.
(C) The graph of the model (Fig. 2B) does not pass through any of the points in the
scatter plot, but it comes close to all of them. [Verify this by evaluating equation
(2) at c = 0.5, 0.6, c, 1.] So we can use equation (2) to approximate points
not in Table 1.
c = 0.85 c = 1.2
p ≈ 6,14010.852 - 480 p ≈ 6,14011.22 - 480
= $4,739 = $6,888
A 0.85-carat diamond will cost about $4,739, and a 1.2-carat diamond will cost
about $6,888.
(D) To find the weight of a $4,000 diamond, we solve the following equation for c:
Table 2 Emerald-Shaped
Diamond Prices 6,140c - 480 = 4,000 Add 480 to both sides.
Weight (carats) Price 6,140c = 4,480 Divide both sides by 6,140.
0.5 $1,677
4,480
0.6 $2,353 c = ≈ 0.73 Rounded to two decimal places.
0.7 $2,718 6,140
0.8 $3,218 A $4,000 diamond will weigh about 0.73 carat.
0.9 $3,982
1.0 $4,510
Matched Problem 3 Prices for emerald-shaped diamonds from an online
Source: www.tradeshop.com
trader are given in Table 2. Repeat Example 3 for this data with the linear model
p = 5,600c - 1,100
where p is the price of an emerald-shaped diamond weighing c carats.
The model we used in Example 3 was obtained using a technique called linear
regression, and the model is called the regression line. This technique produces a
line that is the best fit for a given data set. (The line of best fit is the line that mini-
mizes the sum of the squares of the vertical distances from the data points to the line.)
Although you can find a linear regression line by hand, we prefer to leave the calcula-
tions to a graphing calculator or a computer. Don’t be concerned if you don’t have
either of these electronic devices. We will supply the regression model in most of the
applications we discuss, as we did in Example 3.

As stated previously, we used linear regression to produce the model in Example 3. If you
have a graphing calculator that supports linear regression, then you can find this model.
The linear regression process varies greatly from one calculator to another. Consult the
user’s manual for the details of linear regression. The screens in Figure 3 are related to the
construction of the model in Example 3 on a Texas Instruments TI-84 Plus CE.
(A) Produce similar screens on your graphing calculator.
(B) Do the same for Matched Problem 3.
8000
0
0 1.5
(A) Entering the data (B) Finding the model (C) Graphing the data and the
model
Figure 3 Linear regression on a graphing calculator
In Example 3, we used the regression model to approximate points that were

not given in Table 1 but would fit between points in the table. This process is called
interpolation. In the next example, we use a regression model to approximate points
outside the given data set. This process is called extrapolation, and the approxima-
tions are often referred to as predictions.
EXAMPLE 4 Atmospheric Concentration of Carbon Dioxide Table 3 contains information

about the concentration of carbon dioxide (CO2) in the atmosphere. The linear
regression model for the data is
Table 3 Atmospheric Concentration
of CO2 (parts per million) C = 360 + 2.04t
1995 2000 2005 2010 2015 where C is the concentration (in parts per million) of carbon dioxide and t is the
361 370 380 390 402 time in years with t = 0 corresponding to the year 1995.
Source: National Oceanic and Atmospheric
Administration
(A) Interpret the slope of the regression line as a rate of change.
(B) Use the regression model to predict the concentration of CO2 in the atmosphere
in 2025.
SOLUTION
(A) The slope m = 2.04 is the rate of change of concentration of CO2 with respect
to time. Since the slope is positive, the concentration of CO2 is increasing at a
rate of 2.04 parts per million per year.
(B) If t = 30, then
C = 360 + 2.041302 ≈ 421
So the model predicts that the atmospheric concentration of CO2 will be ap-
proximately 421 parts per million in 2025.
Matched Problem 4 Using the model of Example 4, estimate the concentra-

tion of carbon dioxide in the atmosphere in the year 1990.
Forest managers estimate growth, volume, yield, and forest potential. One com-
mon measure is the diameter of a tree at breast height (Dbh), which is defined as the
diameter of the tree at a point 4.5 feet above the ground on the uphill side of the tree.
Example 5 uses Dbh to estimate the height of balsam fir trees.
EXAMPLE 5 Forestry A linear regression model for the height of balsam fir trees is
h = 3.8d + 18.73
where d is Dbh in inches and h is the height in feet.
(A) Interpret the slope of this model.
(B) What is the effect of a 1-inch increase in Dbh?
(C) Estimate the height of a balsam fir with a Dbh of 8 inches. Round your answer
to the nearest foot.
(D) Estimate the Dbh of a balsam fir that is 30 feet tall. Round your answer to the
nearest inch.
SOLUTION
(A) The rate of change of height with respect to breast height diameter is 3.8 feet
per inch.
(B) Height increases by 3.8 feet.
(C) We must find h when d = 8:
h = 3.8d + 18.73 Substitute d = 8.
h = 3.8182 + 18.73 Evaluate.
h = 49.13 ≈ 49 ft
(D) We must find d when h = 30:
h = 3.8d + 18.73 Substitute h = 30.
30 = 3.8d + 18.73 Subtract 18.73 from both sides.
11.27 = 3.8d Divide both sides by 3.8.
11.27
d = ≈ 3 in.
3.8
The data used to produce the regression model in Example 5 are from the Jack
Haggerty Forest at Lakehead University in Canada (Table 4). We used the popular
Table 4 Height and Diameter of the Balsam Fir

Dbh Height Dbh Height Dbh Height Dbh Height
(in.) (ft) (in.) (ft) (in.) (ft) (in.) (ft)
6.5 51.8 6.4 44.0 3.1 19.7 4.6 26.6
8.6 50.9 4.4 46.9 7.1 55.8 4.8 33.1
5.7 49.2 6.5 52.2 6.3 32.8 3.1 28.5
4.9 46.3 4.1 46.9 2.4 26.2 3.2 29.2
6.4 44.3 8.8 51.2 2.5 29.5 5.0 34.1
4.1 46.9 5.0 36.7 6.9 45.9 3.0 28.2
1.7 13.1 4.9 34.1 2.4 32.8 4.8 33.8
1.8 19.0 3.8 32.2 4.3 39.4 4.4 35.4
3.2 20.0 5.5 49.2 7.3 36.7 11.3 55.4
5.1 46.6 6.3 39.4 10.9 51.5 3.7 32.2
Source: Jack Haggerty Forest, Lakehead University, Canada
spreadsheet Excel to produce a scatter plot of the data in Table 4 and to find the
regression model (Fig. 4).
70.0
y 5 3.80x 1 18.73
60.0
50.0
Height (ft)
40.0
30.0
20.0
10.0
0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Dbh (in.)
Figure 4 Linear regression with a spreadsheet
Matched Problem 5 Figure 5 shows the scatter plot for white spruce trees in
the Jack Haggerty Forest at Lakehead University in Canada. A regression model
produced by a spreadsheet (Fig. 5), after rounding, is
h = 1.8d + 34
where d is Dbh in inches and h is the height in feet.
80.0
y 5 1.80x 1 33.98
70.0
60.0
Height (ft)
50.0
40.0
30.0
20.0
10.0
0.0
0.0 5.0 10.0 15.0 20.0 25.0
Dbh (in.)
Figure 5 Linear regression for white spruce trees
(A) Interpret the slope of this model.

(B) What is the effect of a 1-inch increase in Dbh?
(C) Estimate the height of a white spruce with a Dbh of 10 inches. Round your
answer to the nearest foot.
(D) Estimate the Dbh of a white spruce that is 65 feet tall. Round your answer to
the nearest inch.
Exercises A.3
body weight (in kilograms) and h for height over 5 ft
Applications (in inches).
1. Ideal weight. Dr. J. D. Robinson published the following (B) Interpret the slope of the model.
estimate of the ideal body weight of a woman: (C) If a woman is 5=4== tall, what does the model predict her
49 kg + 1.7 kg for each inch over 5 ft weight to be?
(A) Find a linear model for Robinson’s estimate of (D) If a woman weighs 60 kg, what does the model predict
the ideal weight of a woman using w for ideal her height to be?
2. Ideal weight. Dr. J. D. Robinson also published the follow- at 331 m>sec at 0°C and at 343 m>sec at 20°C, construct
ing estimate of the ideal body weight of a man: a linear model relating the speed of sound (s) and the air
temperature (t). Interpret the slope of this model. (Source:
52 kg + 1.9 kg for each inch over 5 ft
Engineering Toolbox)
(A) Find a linear model for Robinson’s estimate of the ideal
8. Speed of sound. The speed of sound through sea water
weight of a man using w for ideal body weight (in kilo-
is linearly related to the temperature of the water. If sound
grams) and h for height over 5 ft (in inches).
travels at 1,403 m>sec at 0°C and at 1,481 m>sec at 20°C,
(B) Interpret the slope of the model. construct a linear model relating the speed of sound (s) and
the air temperature (t). Interpret the slope of this model.
(C) If a man is 5′8″ tall, what does the model predict his (Source: Engineering Toolbox)
weight to be?
9. Energy production. Table 5 lists U.S. fossil fuel production
(D) If a man weighs 70 kg, what does the model predict his as a percentage of total energy production for selected years.
height to be? A linear regression model for this data is
3. Underwater pressure. At sea level, the weight of the atmo- y = - 0.19x + 83.75
sphere exerts a pressure of 14.7 pounds per square inch, com-
where x represents years since 1985 and y represents the cor-
monly referred to as 1 atmosphere of pressure. As an object
responding percentage of total energy production.
descends in water, pressure P and depth d are linearly related.
In salt water, the pressure at a depth of 33 ft is 2 atms, or 29.4
pounds per square inch. Table 5 U.S. Fossil Fuel Production
(A) Find a linear model that relates pressure P (in pounds per Year Production (%)
square inch) to depth d (in feet). 1985 85
(B) Interpret the slope of the model. 1990 83
1995 81
(C) Find the pressure at a depth of 50 ft. 2000 80
(D) Find the depth at which the pressure is 4 atms. 2005 79
2010 78
4. Underwater pressure. Refer to Problem 3. In fresh water, 2015 80
the pressure at a depth of 34 ft is 2 atms, or 29.4 pounds per
Source: Energy Information Administration
square inch.
(A) Find a linear model that relates pressure P (in pounds per (A) Draw a scatter plot of the data and a graph of the model
square inch) to depth d (in feet). on the same axes.
(B) Interpret the slope of the model. (B) Interpret the slope of the model.
(C) Find the pressure at a depth of 50 ft. (C) Use the model to predict fossil fuel production in 2025.
(D) Find the depth at which the pressure is 4 atms. (D) Use the model to estimate the first year for which fos-
5. Rate of descent—Parachutes. At low altitudes, the altitude sil fuel production is less than 70% of total energy
of a parachutist and time in the air are linearly related. A production.
jump at 2,880 ft using the U.S. Army’s T-10 parachute system 10. Energy consumption. Table 6 lists U.S. fossil fuel consump-
lasts 120 secs. tion as a percentage of total energy consumption for selected
(A) Find a linear model relating altitude a (in feet) and time years. A linear regression model for this data is
in the air t (in seconds). y = - 0.14x + 86.18
(B) Find the rate of descent for a T-10 system. where x represents years since 1985 and y represents the cor-
responding percentage of fossil fuel consumption.
(C) Find the speed of the parachutist at landing.
6. Rate of descent—Parachutes. The U.S Army is considering
a new parachute, the Advanced Tactical Parachute System Table 6 U.S. Fossil Fuel Consumption
(ATPS). A jump at 2,880 ft using the ATPS system lasts 180 Year Consumption (%)
secs.
1985 86
(A) Find a linear model relating altitude a (in feet) and time 1990 85
in the air t (in seconds). 1995 85
2000 84
(B) Find the rate of descent for an ATPS system parachute.
2005 85
(C) Find the speed of the parachutist at landing. 2010 83
7. Speed of sound. The speed of sound through air is lin- 2015 81
early related to the temperature of the air. If sound travels Source: Energy Information Administration
(A) Draw a scatter plot of the data and a graph of the model Table 8 Licensed Drivers in 2014
on the same axes.
State Population Licensed Drivers
(B) Interpret the slope of the model. Alaska 0.74 0.53
(C) Use the model to predict fossil fuel consumption in Delaware 0.94 0.73
2025. Montana 1.00 0.77
North Dakota 0.74 0.53
(D) Use the model to estimate the first year for which fos- South Dakota 0.85 0.61
sil fuel consumption is less than 80% of total energy
Vermont 0.63 0.55
consumption.
Wyoming 0.58 0.42
11. Cigarette smoking. The data in Table 7 shows that the Source: Bureau of Transportation Statistics
percentage of female cigarette smokers in the United States
declined from 22.1% in 1997 to 13.6% in 2015. (B) If the population of Hawaii in 2014 was about 1.4 mil-
lion, use the model to estimate the number of licensed
drivers in Hawaii in 2014 to the nearest thousand.
Table 7 Percentage of Smoking Prevalence
(C) If the number of licensed drivers in Maine in 2014 was
among U.S. Adults
about 1,019,000, use the model to estimate the popula-
Year Males (%) Females (%) tion of Maine in 2014 to the nearest thousand.
1997 27.6 22.1 14. Licensed drivers. Table 9 contains the state population and
2000 25.7 21.0 the number of licensed drivers in the state (both in millions)
2003 24.1 19.2 for the most populous states in 2014. The regression model
2006 23.9 18.0 for this data is
2010 21.5 17.3
y = 0.62x + 0.29
2015 16.7 13.6
Source: Centers for Disease Control where x is the state population (in millions) and y is the num-
ber of licensed drivers (in millions) in the state.
(A) Applying linear regression to the data for females in Table 9 Licensed Drivers in 2014
Table 7 produces the model State Population Licensed Drivers
f = - 0.45t + 22.20 California 39 25
where f is percentage of female smokers and t is time Florida 20 14
in years since 1997. Draw a scatter plot of the female Illinois 13 8
smoker data and a graph of the regression model on the New York 20 11
same axes. Ohio 12 8
(B) Estimate the first year in which the percentage of female Pennsylvania 13 9
smokers is less than 10%. Texas 27 16
Source: Bureau of Transportation Statistics
12. Cigarette smoking. The data in Table 7 shows that the
percentage of male cigarette smokers in the United States (A) Draw a scatter plot of the data and a graph of the model
declined from 27.6% in 1997 to 16.7% in 2015. on the same axes.
(A) Applying linear regression to the data for males in Table 7 (B) If the population of Michigan in 2014 was about 9.9
produces the model million, use the model to estimate the number of licensed
m = - 0.56t + 27.82 drivers in Michigan in 2014 to the nearest thousand.
where m is percentage of male smokers and t is time in (C) If the number of licensed drivers in Georgia in 2014 was
years since 1997. Draw a scatter plot of the male smoker about 6.7 million, use the model to estimate the popula-
data and a graph of the regression model. tion of Georgia in 2014 to the nearest thousand.
(B) Estimate the first year in which the percentage of male
15. Net sales. A linear regression model for the net sales data in
smokers is less than 10%.
Table 10 is
13. Licensed drivers. Table 8 contains the state population and
S = 15.85t + 250.1
the number of licensed drivers in the state (both in millions)
for the states with population under 1 million in 2014. The where S is net sales and t is time since 2000 in years.
regression model for this data is Table 10 Walmart Stores, Inc.
y = 0.75x Billions of
where x is the state population (in millions) and y is the num- U.S. Dollars 2008 2009 2010 2011 2012 2015
ber of licensed drivers (in millions) in the state. Net sales 374 401 405 419 444 488
(A) Draw a scatter plot of the data and a graph of the model Operating income 21.9 22.8 24.0 25.5 26.6 27.3
on the same axes. Source: Walmart Stores, Inc.
(A) Draw a scatter plot of the data and a graph of the model (C) Use the model to estimate the percentage of propylene
on the same axes. glycol in a solution that freezes at 15°F.
(B) Predict Walmart’s net sales for 2026. 19. Forestry. The figure contains a scatter plot of 100 data
points for black spruce trees and the linear regression model
16. Operating income. A linear regression model for the oper-
for this data.
ating income data in Table 10 is
(A) Interpret the slope of the model.
I = 0.82t + 15.84
where I is operating income and t is time since 2000 in years. (B) What is the effect of a 1-in. increase in Dbh?
(A) Draw a scatter plot of the data and a graph of the model (C) Estimate the height of a black spruce with a Dbh of
on the same axes. 15 in. Round your answer to the nearest foot.
(B) Predict Walmart’s annual operating income for 2026. (D) Estimate the Dbh of a black spruce that is 25 ft tall.
Round your answer to the nearest inch.
17. Freezing temperature. Ethylene glycol and propylene glycol
are liquids used in antifreeze and deicing solutions. Ethylene
glycol is listed as a hazardous chemical by the Environmental
Protection Agency, while propylene glycol is generally 45.0
regarded as safe. Table 11 lists the freezing temperature for 40.0
y 5 1.37x 2 2.58
various concentrations (as a percentage of total weight) of 35.0
each chemical in a solution used to deice airplanes. A linear 30.0
Height (ft)
regression model for the ethylene glycol data in Table 11 is 25.0
20.0
E = - 0.55T + 31 15.0
where E is the percentage of ethylene glycol in the deic- 10.0
ing solution and T is the temperature at which the solution 5.0
freezes. 0.0
0.0 5.0 10.0 15.0 20.0 25.0
Dbh (in.)
Table 11 Freezing Temperatures
Freezing black spruce
Temperature Ethylene Propylene Source: Lakehead University
(°F) Glycol (% Wt.) Glycol (% Wt.)
-50 56 58
20. Forestry. The figure contains a scatter plot of 100 data
-40 53 55
points for black walnut trees and the linear regression model
-30 49 52
for this data.
-20 45 48
-10 40 43 (A) Interpret the slope of the model.
0 33 36
10 25 29
20 16 19 45.0
Source: T. Labuza, University of Minnesota 40.0
y 5 1.66x 2 5.14
35.0
(A) Draw a scatter plot of the data and a graph of the model 30.0
Height (ft)
on the same axes. 25.0

20.0
(B) Use the model to estimate the freezing temperature to the
15.0
nearest degree of a solution that is 30% ethylene glycol.
10.0
(C) Use the model to estimate the percentage of ethylene 5.0
glycol in a solution that freezes at 15°F. 0.0
0.0 5.0 10.0 15.0 20.0 25.0
18. Freezing temperature. A linear regression model for the Dbh (in.)
propylene glycol data in Table 11 is
black walnut
P = - 0.54T + 34 Source: Kagen Research
where P is the percentage of propylene glycol in the deicing
solution and T is the temperature at which the solution freezes. (B) What is the effect of a 1-in. increase in Dbh?
(A) Draw a scatter plot of the data and a graph of the model (C) Estimate the height of a black walnut with a Dbh of
on the same axes. 12 in. Round your answer to the nearest foot.
(B) Use the model to estimate the freezing temperature to the (D) Estimate the Dbh of a black walnut that is 25 ft tall.
nearest degree of a solution that is 30% propylene glycol. Round your answer to the nearest inch.
21. Undergraduate enrollment. Table 12 lists fall undergraduate

enrollment by gender in U.S. degree-granting institutions.
The figure contains a scatter plot and regression line for each
data set, where x represents years since 1980 and y represents
enrollment (in millions).
(A) Interpret the slope of each model.
(B) Use the regression models to predict the male and
female undergraduate enrollments in 2025.
(C) Use the regression models to estimate the first year in (A) (B)
which female undergraduate enrollment will exceed Figure for 22
male undergraduate enrollment by at least 3 million.
Problems 23–28 require a graphing calculator or a computer that
Table 12 Fall Undergraduate can calculate the linear regression line for a given data set.
Enrollment (in millions)
23. Climate. Find a linear regression model for the data on aver-
Year Male Female age annual temperature in Table 14, where x is years since
1980 5.00 5.47 1960 and y is temperature (in °F). (Round regression coeffi-
1990 5.38 6.58 cients to three decimal places). Use the model to estimate the
2000 5.78 7.38 average annual temperature in the contiguous United States
2010 7.84 10.25 in 2025.
2014 7.59 9.71
Table 14 Climate Data for the Contiguous
Source: National Center for Education
Statistics
United States
Average Annual Average Annual
Year Temperature (°F) Precipitation (in)
1965 51.69 29.80
1975 51.50 33.03
1985 51.30 29.97
1995 52.65 32.69
2005 53.64 30.08
2015 54.40 34.59
Source: National Oceanic and Atmospheric Administration
(A) (B)
24. Climate. Find a linear regression model for the data on
Figure for 21 average annual precipitation in Table 14, where x is years
22. Graduate enrollment. Table 13 lists fall graduate enroll- since 1960 and y is precipitation (in inches). (Round regres-
ment by gender in U.S. degree-granting institutions. The sion coefficients to three decimal places). Use the model to
figure contains a scatter plot and regression line for each data estimate the average annual precipitation in the contiguous
set, where x represents years since 1980 and y represents United States in 2025.
enrollment (in millions). 25. Olympic Games. Find a linear regression model for the men’s
(A) Interpret the slope of each model. 100-meter freestyle data given in Table 15, where x is years
since 1990 and y is winning time (in seconds). Do the same for
(B) Use the regression models to predict the male and the women’s 100-meter freestyle data. (Round regression coef-
female graduate enrollments in 2025. ficients to three decimal places.) Do these models indicate that
(C) Use the regression models to estimate the first year in the women will eventually catch up with the men?
which female graduate enrollment will exceed male
graduate enrollment by at least 1 million. Table 15 Winning Times in Olympic Swimming Events
Table 13 Fall Graduate Enrollment 100-Meter Freestyle 200-Meter Backstroke

(in millions) Men Women Men Women
Year Male Female 1992 49.02 54.65 1:58.47 2:07.06
1996 48.74 54.50 1.58.54 2:07.83
1980 0.87 0.75
2000 48.30 53.83 1:56.76 2:08.16
1990 0.90 0.96
2004 48.17 53.84 1:54.76 2:09.16
2000 0.94 1.21
2008 47.21 53.12 1:53.94 2:05.24
2010 1.21 1.73
2012 47.52 53.00 1:53.41 2:04.06
2014 1.21 1.70
2016 47.58 52.70 1:53:62 2:05.99
Source: National Center for Education
Statistics Source: www.infoplease.com
26. Olympic Games. Find a linear regression model for the Answers to Matched Problems
men’s 200-meter backstroke data given in Table 15, where
1. (A) The rate of change of BSA with respect to weight is
x is years since 1990 and y is winning time (in seconds). Do
28.55 square inches per pound.
the same for the women’s 200-meter backstroke data. (Round
regression coefficients to three decimal places.) Do these (B) Increasing w by 1 pound increases a by 28.55 square
models indicate that the women will eventually catch up with inches.
the men? 2. (A) a = - 36t + 2,880
27. Supply and demand. Table 16 contains price–supply (B) 36 ft/sec
data and price–demand data for corn. Find a linear regression 3. (A) p
model for the price–supply data where x is supply (in billions
of bushels) and y is price (in dollars). Do the same for the
Price (dollars)
6000
price–demand data. (Round regression coefficients to two 4000
decimal places.) Find the equilibrium price for corn.
2000
Table 16 Supply and Demand for U.S. Corn c
0 0.5 1.0 1.5
Supply Demand Carats
Price ($/bu) (billion bu) Price ($/bu) (billion bu)
2.15 6.29 2.07 9.78 p
2.29 7.27 2.15 9.35
Price (dollars)
6000
2.36 7.53 2.22 8.47
2.48 7.93 2.34 8.12 4000
2.47 8.12 2.39 7.76 2000
2.55 8.24 2.47 6.98
c
Source: www.usda.gov/nass/pubs/histdata.htm 0 0.5 1.0 1.5
Carats
28. Supply and demand. Table 17 contains price–supply
data and price–demand data for soybeans. Find a linear (B) The rate of change of the price of a diamond with respect
regression model for the price–supply data where x is to its weight is $5,600. Increasing the weight by one
supply (in billions of bushels) and y is price (in dollars). carat will increase the price by about $5,600.
Do the same for the price–demand data. (Round regression (C) $3,660; $5,620
coefficients to two decimal places.) Find the equilibrium (D) 0.91 carat
price for soybeans.
4. Approximately 350 parts per million.
Table 17 Supply and Demand for U.S. Soybeans 5. (A) The slope is 1.8, so the rate of change of height with
respect to breast height diameter is 1.8 feet per inch.
Supply Demand
Price ($/bu) (billion bu) Price ($/bu) (billion bu) (B) Height increases by 1.8 feet.
5.15 1.55 4.93 2.60 (C) 52 ft
5.79 1.86 5.48 2.40 (D) 17 in.
5.88 1.94 5.71 2.18
6.07 2.08 6.07 2.05
6.15 2.15 6.40 1.95
6.25 2.27 6.66 1.85
Source: www.usda.gov/nass/pubs/histdata.htm
Appendix
B
B.1 Real Numbers
B.2 Operations on Polynomials

Basic Algebra
Review
Appendix B reviews some important basic algebra concepts usually studied
in earlier courses. The material may be studied systematically before begin-
ning the rest of the book or reviewed as needed.
B.3 Factoring Polynomials
B.4 Operations on Rational

Expressions
B.5 Integer Exponents and
Scientific Notation
B.6 Rational Exponents and
Radicals
B.7 Quadratic Equations
B.1 Real Numbers

■■ Set of Real Numbers The rules for manipulating and reasoning with symbols in algebra depend, in large mea-
■■ Real Number Line
sure, on properties of the real numbers. In this section we look at some of the important
properties of this number system. To make our discussions here and elsewhere in the book
■■ Basic Real Number Properties clearer and more precise, we occasionally make use of simple set concepts and notation.
■■ Further Properties
■■ Fraction Properties
Set of Real Numbers
Informally, a real number is any number that has a decimal representation. The
decimal representation may be terminating or repeating or neither. The decimal rep-
resentation 4.713 516 94 is terminating (the space after every third decimal place is
used to help keep track of the number of decimal places). The decimal representation
5.254 747 is repeating (the overbar indicates that the block “47” repeats indefinitely).
The decimal representation 3.141 592 653 c of the number p, the ratio of the cir-
cumference to the diameter of a circle, is neither terminating nor repeating. Table 1
describes the set of real numbers and some of its important subsets. Figure 1 illustrates
how these sets of numbers are related.
Table 1 Set of Real Numbers
Symbol Name Description Examples
N Natural numbers Counting numbers (also called positive integers) 1, 2, 3, c
Z Integers Natural numbers, their negatives, and 0 c , -2, -1, 0, 1, 2, c
Q Rational Numbers that can be represented as a/b, where a -4, 0, 1, 25, -3 2
3.67, -0.333, 5.272 727
5 , 3,
numbers and b are integers and b ∙ 0; decimal represen-
tations are repeating or terminating
I Irrational Numbers that can be represented as nonrepeating 3
22, p, 2 7, 1.414 213 c , 2.718 281 82 . .
numbers and nonterminating decimal numbers
R Real numbers Rational and irrational numbers
567
568 APPENDIX B Basic Algebra Review
Natural
numbers (N)
Zero Integers (Z)
Rational
numbers (Q)
Negatives Noninteger Real
of natural ratios of numbers (R)
numbers integers
Irrational
numbers (I)
Figure 1 Real numbers and important subsets
The set of integers contains all the natural numbers and something else—their
negatives and 0. The set of rational numbers contains all the integers and something
else—noninteger ratios of integers. And the set of real numbers contains all the ratio-
nal numbers and something else—irrational numbers.
Real Number Line

A one-to-one correspondence exists between the set of real numbers and the set of points
on a line. That is, each real number corresponds to exactly one point, and each point
corresponds to exactly one real number. A line with a real number associated with each
point, and vice versa, as shown in Figure 2, is called a real number line, or simply a
Origin
real line. Each number associated with a point is called the coordinate of the point.
210 25 0 5 10 The point with coordinate 0 is called the origin. The arrow on the right end of
17 p the line indicates a positive direction. The coordinates of all points to the right of the
28.34 22 Ï73
4
origin are called positive real numbers, and those to the left of the origin are called
Figure 2 Real number line negative real numbers. The real number 0 is neither positive nor negative.
Basic Real Number Properties

We now take a look at some of the basic properties of the real number system that
enable us to convert algebraic expressions into equivalent forms.
SUMMARY Basic Properties of the Set of Real Numbers

Let a, b, and c be arbitrary elements in the set of real numbers R.
Addition Properties
Associative: 1a + b2 + c = a + 1b + c2
Commutative: a + b = b + a
Identity: 0 is the additive identity; that is, 0 + a = a + 0 = a for all a in
R, and 0 is the only element in R with this property.
Inverse: For each a in R, - a, is its unique additive inverse; that is,
a + 1 - a2 = 1 - a2 + a = 0 and - a is the only element in R
relative to a with this property.
Multiplication Properties
Associative: 1ab2c = a1bc2
Commutative: ab = ba
Identity: 1 is the multiplicative identity; that is, 112a = a112 = a for all a
in R, and 1 is the only element in R with this property.
Inverse: For each a in R, a ∙ 0, 1>a is its unique multiplicative inverse;
that is, a11>a2 = 11>a2a = 1, and 1>a is the only element in R
relative to a with this property.
Distributive Properties
a1b + c2 = ab + ac 1a + b2c = ac + bc
SECTION B.1 Real Numbers 569
You are already familiar with the commutative properties for addition and mul-
tiplication. They indicate that the order in which the addition or multiplication of two
numbers is performed does not matter. For example,
7 + 2 = 2 + 7 and 3 # 5 = 5 # 3
Is there a commutative property relative to subtraction or division? That is, does
a - b = b - a or does a , b = b , a for all real numbers a and b (division by 0
excluded)? The answer is no, since, for example,
8 - 6 ∙ 6 - 8 and 10 , 5 ∙ 5 , 10
When computing
3 + 2 + 6 or 3 # 2 # 6
why don’t we need parentheses to indicate which two numbers are to be added or
multiplied first? The answer is to be found in the associative properties. These prop-
erties allow us to write
13 + 22 + 6 = 3 + 12 + 62 and 13 # 22 # 6 = 3 # 12 # 62
so it does not matter how we group numbers relative to either operation. Is there an
associative property for subtraction or division? The answer is no, since, for example,
112 - 62 - 2 ∙ 12 - 16 - 22 and 112 , 62 , 2 ∙ 12 , 16 , 22
Evaluate each side of each equation to see why.
What number added to a given number will give that number back again? What
number times a given number will give that number back again? The answers are 0
and 1, respectively. Because of this, 0 and 1 are called the identity elements for the
real numbers. Hence, for any real numbers a and b,
0 + 5 = 5 and 1a + b2 + 0 = a + b
1 # 4 = 4 and 1a + b2 # 1 = a + b
We now consider inverses. For each real number a, there is a unique real number
- a such that a + 1 - a2 = 0. The number - a is called the additive inverse of a, or
the negative of a. For example, the additive inverse (or negative) of 7 is - 7, since
7 + 1 - 72 = 0. The additive inverse (or negative) of - 7 is - 1 - 72 = 7, since
- 7 + 3- 1 - 724 = 0.
CONCEPTUAL INSIGHT
Do not confuse negation with the sign of a number. If a is a real number, - a is
the negative of a and may be positive or negative. Specifically, if a is negative,
then - a is positive and if a is positive, then - a is negative.
For each nonzero real number a, there is a unique real number 1>a such that
a11>a2 = 1. The number 1>a is called the multiplicative inverse of a, or the
reciprocal of a. For example, the multiplicative inverse (or reciprocal) of 4 is 14, since
4 1 14 2 = 1. (Also note that 4 is the multiplicative inverse of 14.) The number 0 has no
multiplicative inverse.
We now turn to the distributive properties, which involve both multiplication
and addition. Consider the following two computations:
513 + 42 = 5 # 7 = 35 5 # 3 + 5 # 4 = 15 + 20 = 35
Thus,
513 + 42 = 5 # 3 + 5 # 4
and we say that multiplication by 5 distributes over the sum 13 + 42. In general,
multiplication distributes over addition in the real number system. Two more
illustrations are
91m + n2 = 9m + 9n 17 + 22u = 7u + 2u
EXAMPLE 1 Real Number Properties State the real number property that justifies the indi-
cated statement.
Statement Property Illustrated
(A) x1y + z2 = 1y + z2x Commutative ( # )
(B) 512y2 = 15 # 22y Associative ( # )
(C) 2 + 1y + 72 = 2 + 17 + y2 Commutative ( + )
(D) 4z + 6z = 14 + 62z Distributive
(E) If m + n = 0, then n = -m. Inverse ( + )
MATCHED PROBLEM 1 State the real number property that justifies the indicated
statement.
(A) 8 + 13 + y2 = 18 + 32 + y
(B) 1x + y2 + z = z + 1x + y2
(C) 1a + b21x + y2 = a1x + y2 + b1x + y2
(D) 5xy + 0 = 5xy
(E) If xy = 1, x ∙ 0, then y = 1>x.
Further Properties
Subtraction and division can be defined in terms of addition and multiplication,
respectively:
DEFINITION Subtraction and Division

For all real numbers a and b,
Subtraction: a - b = a + 1 - b2 7 - 1 - 52 = 7 + 3 - 1 - 52 4
= 7 + 5 = 12
1 1 9
Division: a , b = a a b, b ∙ 0 9 , 4 = 9a b =
b 4 4
To subtract b from a, add the negative (the additive inverse) of b to a. To divide a

by b, multiply a by the reciprocal (the multiplicative inverse) of b. Note that division by
0 is not defined, since 0 does not have a reciprocal. 0 can never be used as a divisor!
The following properties of negatives can be proved using the preceding assumed
properties and definitions.
THEOREM 1 Negative Properties

-a a a
1. - 1 - a2 = a 5. = - = ,b ∙ 0
2. 1 - a2b = - 1ab2
b b -b
-a -a a a
= a1 - b2 = - ab 6. = - = - = ,b ∙ 0
3. 1 - a21 - b2 = ab
-b b -b b
4. 1 - 12a = - a
We now state two important properties involving 0.
THEOREM 2 Zero Properties

1. a # 0 = 0 0 # 0 = 0 1 - 352 102 = 0
2. ab = 0 if and only if a = 0 or b = 0
If 13x + 22 1x - 72 = 0, then either 3x + 2 = 0 or x - 7 = 0.
SECTION B.1 Real Numbers 571
Fraction Properties
Recall that the quotient a , b1b ∙ 02 written in the form a>b is called a fraction.
The quantity a is called the numerator, and the quantity b is called the denominator.
THEOREM 3 Fraction Properties

For all real numbers a, b, c, d, and k (division by 0 excluded):
a c
since 4 # 9 = 6 # 6
4 6
1. = if and only if ad = bc =
b d 6 9
3. # = = #
ka a a c ac a c a d
2. = 4. ,
kb b b d bd b d b c
7#3 3#7 3#7
= #
3 2 5 2 7
7#5
= = # ,
5 5 8 5 8 3 7 3 5
a c a + c a c a - c a c ad + bc
5. + = 6. - = 7. + =
b b b b b b b d bd
3 5 3 + 5 7 3 7 - 3 2 3 2#5 + 3#3
3#5
+ = - = + =
6 6 6 8 8 8 3 5
A fraction is a quotient, not just a pair of numbers. So if a and b are real numbers
with b ∙ 0, then ab corresponds to a point on the real number line. For example, 17 2
corresponds to the point halfway between 16 18 21
2 = 8 and 2 = 9. Similarly, - 5 corre-
sponds to the point that is 15 unit to the left of - 4.
EXAMPLE 2 Estimation Round 22

7 +
18
19 to the nearest integer.
22 18
SOLUTION Note that a calculator is not required: 7 is a little greater than 3, and 19
is a little less than 1. Therefore the sum, rounded to the nearest integer, is 4.
6
MATCHED PROBLEM 2 Round 93 to the nearest integer.
Fractions with denominator 100 are called percentages. They are used so often
that they have their own notation:
3 7.5 110
= 3% = 7.5% = 110%
100 100 100
So 3% is equivalent to 0.03, 7.5% is equivalent to 0.075, and so on.
EXAMPLE 3 State Sales Tax Find the sales tax that is owed on a purchase of $947.69 if the tax
rate is 6.5%.
SOLUTION 6.5%1$947.692 = 0.0651947.692 = $61.60
MATCHED PROBLEM 3 You intend to give a 20% tip, rounded to the nearest dol-
lar, on a restaurant bill of $78.47. How much is the tip?
Exercises B.1
All variables represent real numbers. 3. Associative property 1 + 2: 3 + 17 + y2 = ?
A In Problems 1–6, replace each question mark with an appropriate 4. Associative property 1 # 2: x1yz2 = ?
expression that will illustrate the use of the indicated real number 5. Identity property 1 # 2: 11u + v2 = ?
property.
6. Identity property 1 + 2: 0 + 9m = ?
1. Commutative property 1 # 2: uv = ?
2. Commutative property 1 + 2: x + 7 = ?
In Problems 7–26, indicate true (T) or false (F). 33. Given the sets of numbers N (natural numbers), Z (integers),
7. 518m2 = 15 # 82m
Q (rational numbers), and R (real numbers), indicate to which
set(s) each of the following numbers belongs:
8. a + cb = a + bc -5
(A) 8 (B) 12 (C) - 1.414 (D) 2
9. 5x + 7x = 15 + 72x
34. Given the sets of numbers N, Z, Q, and R (see Problem 33), in-
10. uv1w + x2 = uvw + uvx dicate to which set(s) each of the following numbers belongs:
11. - 21 - a2 12x - y2 = 2a1 - 4x + y2 (A) - 3 (B) 3.14 (C) p (D) 2
3
1 35. Indicate true (T) or false (F), and for each false statement find
12. 8 , 1 - 52 = 8a b
-5 real number replacements for a, b, and c that will provide a
13. 1x + 32 + 2x = 2x + 1x + 32 counterexample. For all real numbers a, b, and c,
(A) a1b - c2 = ab - c
x 5y 15y 2
14. , = 2 (B) 1a - b2 - c = a - 1b - c2
3y x x
2x 2x (C) a1bc2 = 1ab2c
15. = -
- 1x + 32 x + 3 (D) 1a , b2 , c = a , 1b , c2
2x 2x 36. Indicate true (T) or false (F), and for each false statement
16. - =
- 1x - 32 x - 3 find real number replacements for a and b that will provide a
counterexample. For all real numbers a and b,
1
17. 1 - 32 a b = 1 (A) a + b = b + a
-3
18. 1 - 0.52 + 10.52 = 0 (B) a - b = b - a
(C) ab = ba
19. - x2y 2 = 1 - 12x2y 2
(D) a , b = b , a
20. 3 - 1x + 22 41 - x2 = 1x + 22x
C 37. If c = 0.151515 c , then 100c = 15.1515 c and
a c a + c
21. + = 100c - c = 15.1515 c - 0.151515 c
b d b + d
99c = 15
k 1
22. = 15 5
k + b 1 + b c = =
99 33
23. 1x + 82 1x + 62 = 1x + 82x + 1x + 826
Proceeding similarly, convert the repeating decimal
24. u1u - 2v2 + v1u - 2v2 = 1u + v2 1u - 2v2 0.090909 c into a fraction. (All repeating decimals are
25. If 1x - 22 12x + 32 = 0, then either x - 2 = 0 or rational numbers, and all rational numbers have repeating
2x + 3 = 0. decimal representations.)
26. If either x - 2 = 0 or 2x + 3 = 0, then 38. Repeat Problem 37 for 0.181818 c.

1x - 22 12x + 32 = 0.
Use a calculator to express each number in Problems 39 and
B 27. If uv = 1, does either u or v have to be 1? Explain. 40 as a decimal to the capacity of your calculator. Observe
28. If uv = 0, does either u or v have to be 0? Explain. the repeating decimal representation of the rational numbers
and the nonrepeating decimal representation of the irrational
29. Indicate whether the following are true (T) or false (F): numbers.
(A) All integers are natural numbers. 39. (A) 13
(B) 121 (C) 7
(D) 29
6 16 111
(B) All rational numbers are real numbers. 8 3 11
40. (A) 9 (B) 11 (C) 15 (D) 8
(C) All natural numbers are rational numbers.
In Problems 41–44, without using a calculator, round to the near-
30. Indicate whether the following are true (T) or false (F):
est integer.
(A) All natural numbers are integers.
43 37
41. (A) 13 (B) 19
(B) All real numbers are irrational.
9 12
42. (A) 17 (B) - 25
(C) All rational numbers are real numbers.
7 11 55 7
31. Give an example of a real number that is not a rational 43. (A) 8 + 12 (B) 9 - 55
number. 44. (A) 5
- 18
(B) 13
+ 44
6 19 5 21
32. Give an example of a rational number that is not an
integer.
SECTION B.2 Operations on Polynomials 573
48. Gasoline prices. The price of gas increased 4% in one

Applications week. If the price last week was $4.30 per gallon, what is the
price now? (Round to the nearest cent).
45. Sales tax. Find the tax owed on a purchase of $182.39 if the
state sales tax rate is 9%. (Round to the nearest cent).
46. Sales tax. If you paid $29.86 in tax on a purchase of 1. (A) Associative 1 + 2 (B) Commutative 1 + 2
$533.19, what was the sales tax rate? (Write as a percentage, (C) Distributive (D) Identity 1 + 2
rounded to one decimal place). (E) Inverse 1 # 2
47. Gasoline prices. If the price per gallon of gas jumped from 2. 0 3. $16
$4.25 to $4.37, what was the percentage increase? (Round to
one decimal place).
B.2 Operations on Polynomials

■■ Natural Number Exponents This section covers basic operations on polynomials. Our discussion starts with a
■■ Polynomials
brief review of natural number exponents. Integer and rational exponents and their
properties will be discussed in detail in subsequent sections. (Natural numbers, inte-
■■ Combining Like Terms gers, and rational numbers are important parts of the real number system; see Table 1
■■ Addition and Subtraction and Figure 1 in Appendix B.1.)
■■ Multiplication
Natural Number Exponents
■■ Combined Operations
We define a natural number exponent as follows:
DEFINITION Natural Number Exponent

For n a natural number and b any real number,
bn = b # b # g # b n factors of b
35
= 3#3#3#3#3 5 factors of 3
where n is called the exponent and b is called the base.
Along with this definition, we state the first property of exponents:
THEOREM 1 First Property of Exponents

For any natural numbers m and n, and any real number b:
bm bn = bm + n 12t 4 2 15t 3 2 = 2 # 5t 4 + 3 = 10t 7
Polynomials
Algebraic expressions are formed by using constants and variables and the algebraic
operations of addition, subtraction, multiplication, division, raising to powers, and
taking roots. Special types of algebraic expressions are called polynomials. A poly-
nomial in one variable x is constructed by adding or subtracting constants and terms
of the form axn, where a is a real number and n is a natural number. A polynomial
in two variables x and y is constructed by adding and subtracting constants and
terms of the form axmy n, where a is a real number and m and n are natural numbers.
Polynomials in three and more variables are defined in a similar manner.
Polynomials Not Polynomials
1 x - y
8 0 x x2 + y 2
3x3 - 6x + 7 6x + 3
2x2 - 7xy - 8y 2 9y 3 + 4y 2 - y + 4 2x3 - 2x 2x - 2 - 3x -1
2x - 3y + 2 5 3 2
u - 3u v + 2uv - v 4 4
Polynomial forms are encountered frequently in mathematics. For the efficient

study of polynomials, it is useful to classify them according to their degree. If a term
in a polynomial has only one variable as a factor, then the degree of the term is the
power of the variable. If two or more variables are present in a term as factors, then
the degree of the term is the sum of the powers of the variables. The degree of a
polynomial is the degree of the nonzero term with the highest degree in the polyno-
mial. Any nonzero constant is defined to be a polynomial of degree 0. The number 0
is also a polynomial but is not assigned a degree.
EXAMPLE 1 Degree
(A) The degree of the first term in 5x3 + 13x - 12 is 3, the degree of the second
term is 1, the degree of the third term is 0, and the degree of the whole polyno-
mial is 3 (the same as the degree of the term with the highest degree).
(B) The degree of the first term in 8u3v2 - 17uv2 is 5, the degree of the second
term is 3, and the degree of the whole polynomial is 5.
Matched Problem 1
(A) Given the polynomial 6x5 + 7x3 - 2, what is the degree of the first term? The
second term? The third term? The whole polynomial?
(B) Given the polynomial 2u4v2 - 5uv3, what is the degree of the first term? The
second term? The whole polynomial?
In addition to classifying polynomials by degree, we also call a single-term poly-

nomial a monomial, a two-term polynomial a binomial, and a three-term polynomial
a trinomial.
Combining Like Terms

The concept of coefficient plays a central role in the process of combining like terms.
A constant in a term of a polynomial, including the sign that precedes it, is called the
numerical coefficient, or simply, the coefficient, of the term. If a constant does not
appear, or only a + sign appears, the coefficient is understood to be 1. If only a -
sign appears, the coefficient is understood to be - 1. Given the polynomial
5x4 - x3 - 3x2 + x - 7 = 5x4 + 1 - 12x3 + 1 - 32x2 + 1x + 1 - 72
the coefficient of the first term is 5, the coefficient of the second term is - 1, the
coefficient of the third term is - 3, the coefficient of the fourth term is 1, and the coef-
ficient of the fifth term is - 7.
The following distributive properties are fundamental to the process of combin-
ing like terms.
THEOREM 2 Distributive Properties of Real Numbers

1. a1b + c2 = 1b + c2a = ab + ac
2. a1b - c2 = 1b - c2a = ab - ac
3. a1b + c + g + f2 = ab + ac + g + af
Two terms in a polynomial are called like terms if they have exactly the same
variable factors to the same powers. The numerical coefficients may or may not be
the same. Since constant terms involve no variables, all constant terms are like terms.
If a polynomial contains two or more like terms, these terms can be combined into
a single term by making use of distributive properties. The following example illus-
trates the reasoning behind the process:
3x2y - 5xy 2 + x2y - 2x2y = 3x2y + x2y - 2x2y - 5xy 2
= 13x2y + 1x2y - 2x2y2 - 5xy 2 Note the use
of distributive
= 13 + 1 - 22x2y - 5xy 2 properties.
= 2x2y - 5xy 2
Free use is made of the real number properties discussed in Appendix B.1.
How can we simplify expressions such as 41x - 2y2 - 312x - 7y2? We clear
the expression of parentheses using distributive properties, and combine like terms:
41x - 2y2 - 312x - 7y2 = 4x - 8y - 6x + 21y
= - 2x + 13y
EXAMPLE 2 Removing Parentheses Remove parentheses and simplify:

(A) 213x2 - 2x + 52 + 1x2 + 3x - 72 = 213x2 - 2x + 52 + 11x2 + 3x - 72
= 6x2 - 4x + 10 + x2 + 3x - 7
= 7x2 - x + 3
3 3 2
(B) 1x - 2x - 62 - 12x - x + 2x - 32
= 11x3 - 2x - 62 + 1 - 1212x3 - x2 + 2x - 32 Be careful with the
sign here
= x3 - 2x - 6 - 2x3 + x2 - 2x + 3
= - x3 + x2 - 4x - 3
(C) 33x2 - 12x + 124 - 1x2 - 12 = 33x2 - 2x - 14 - 1x2 - 12
= 3x2 - 2x - 1 - x2 + 1
= 2x2 - 2x
MATCHED PROBLEM 2 Remove parentheses and simplify:
(A) 31u2 - 2v2 2 + 1u2 + 5v2 2
(B) 1m3 - 3m2 + m - 12 - 12m3 - m + 32
(C) 1x3 - 22 - 32x3 - 13x + 424

Addition and subtraction of polynomials can be thought of in terms of removing
parentheses and combining like terms, as illustrated in Example 2. Horizontal and
vertical arrangements are illustrated in the next two examples. You should be able to
work either way, letting the situation dictate your choice.
EXAMPLE 3 Adding Polynomials Add horizontally and vertically:

x4 - 3x3 + x2, - x3 - 2x2 + 3x, and 3x2 - 4x - 5
SOLUTION Add horizontally:
1x4 - 3x3 + x2 2 + 1 - x3 - 2x2 + 3x2 + 13x2 - 4x - 52
= x4 - 3x3 + x2 - x3 - 2x2 + 3x + 3x2 - 4x - 5
= x4 - 4x3 + 2x2 - x - 5
Or vertically, by lining up like terms and adding their coefficients:
x4 - 3x3 + x2
- x3 - 2x2 + 3x
3x2 - 4x - 5
x4 - 4x3 + 2x2 - x - 5
MATCHED PROBLEM 3 Add horizontally and vertically:

3x - 2x - 4x2,
4 3
x3 - 2x2 - 5x, and x2 + 7x - 2
EXAMPLE 4 Subtracting Polynomials Subtract 4x2 - 3x + 5 from x2 - 8, both horizontally

and vertically.
SOLUTION 1x2 - 82 - 14x2 - 3x + 52 or x2 - 8
2 2 2
= x - 8 - 4x + 3x - 5 - 4x + 3x - 5 d Change
2 2 signs and
= - 3x + 3x - 13 - 3x + 3x - 13
add.
MATCHED PROBLEM 4 Subtract 2x2 - 5x + 4 from 5x2 - 6, both horizontally

and vertically.
Multiplication
Multiplication of algebraic expressions involves the extensive use of distributive
properties for real numbers, as well as other real number properties.
EXAMPLE 5 Multiplying Polynomials Multiply: 12x - 3213x2 - 2x + 32

SOLUTION
12x - 3213x2 - 2x + 32 = 2x13x2 - 2x + 32 - 313x2 - 2x + 32

= 6x3 - 4x2 + 6x - 9x2 + 6x - 9
= 6x3 - 13x2 + 12x - 9
Or, using a vertical arrangement,
3x2 - 2x + 3
2x - 3
6x3 - 4x2 + 6x
- 9x2 + 6x - 9
6x3 - 13x2 + 12x - 9
MATCHED PROBLEM 5 Multiply: 12x - 3212x2 + 3x - 22
Thus, to multiply two polynomials, multiply each term of one by each term of
the other, and combine like terms.
Products of binomial factors occur frequently, so it is useful to develop proce-
dures that will enable us to write down their products by inspection. To find the pro-
duct 12x - 1213x + 22 we proceed as follows:
12x - 1213x + 22 = 6x2 + 4x - 3x - 2 The inner and outer products

are like terms, so combine into
= 6x2 + x - 2
a single term.
To speed the process, we do the step in the dashed box mentally.
Products of certain binomial factors occur so frequently that it is useful to learn
formulas for their products. The following formulas are easily verified by multiplying
the factors on the left.
THEOREM 3 Special Products

1. 1a - b21a + b2 = a2 - b2
2. 1a + b2 2 = a2 + 2ab + b2
3. 1a - b2 2 = a2 - 2ab + b2
EXAMPLE 6 Special Products Multiply mentally, where possible.

(A) 12x - 3y215x + 2y2 (B) 13a - 2b213a + 2b2
2
(C) 15x - 32 (D) 1m + 2n2 3
SOLUTION
(A) 12x - 3y215x + 2y2 = 10x2 + 4xy - 15xy - 6y 2
= 10x2 - 11xy - 6y 2
(B) 13a - 2b213a + 2b2 = 13a2 2 - 12b2 2
= 9a2 - 4b2
(C) 15x - 32 2 = 15x2 2 - 215x2132 + 32

= 25x2 - 30x + 9
(D) 1m + 2n2 3 = 1m + 2n2 2 1m + 2n2

= 1m2 + 4mn + 4n2 21m + 2n2
= m2 1m + 2n2 + 4mn1m + 2n2 + 4n2 1m + 2n2
= m3 + 2m2n + 4m2n + 8mn2 + 4mn2 + 8n3
= m3 + 6m2n + 12mn2 + 8n3
MATCHED PROBLEM 6 Multiply mentally, where possible.

(A) 14u - 3v212u + v2 (B) 12xy + 3212xy - 32
(C) 1m + 4n21m - 4n2 (D) 12u - 3v2 2
(E) 12x - y2 3
Combined Operations
We complete this section by considering several examples that use all the operations
just discussed. Note that in simplifying, we usually remove grouping symbols start-
ing from the inside. That is, we remove parentheses 12 first, then brackets 34, and
finally braces 5 6, if present. Also, we observe the following order of operations.
DEFINITION Order of Operations

Multiplication and division precede addition and subtraction, and taking powers
precedes multiplication and division.
2 # 3 + 4 = 6 + 4 = 10, not 2 # 7 = 14
102 100
= = 50, not 52 = 25
2 2
EXAMPLE 7 Combined Operations Perform the indicated operations and simplify:

(A) 3x - 55 - 33x - x13 - x246 = 3x - 55 - 33x - 3x + x2 46
= 3x - 55 - 3x + 9x - 3x2 6
= 3x - 5 + 3x - 9x + 3x2
= 3x2 - 3x - 5
(B) 1x - 2y212x + 3y2 - 12x + y2 2 = 2x2 - xy - 6y 2 - 14x2 + 4xy + y 2 2
= 2x2 - xy - 6y 2 - 4x2 - 4xy - y 2
= - 2x2 - 5xy - 7y 2
MATCHED PROBLEM 7 Perform the indicated operations and simplify:
(A) 2t - 57 - 23t - t14 + t246
(B) 1u - 3v2 2 - 12u - v212u + v2
Exercises B.2
A Problems 1–8 refer to the following polynomials: 37. 1m - 22 2 - 1m - 22 1m + 22
(A) 2x - 3 (B) 2x2 - x + 2 (C) x3 + 2x2 - x + 3 38. 1x - 32 1x + 32 - 1x - 32 2
1. What is the degree of (C)? 39. 1x - 2y2 12x + y2 - 1x + 2y2 12x - y2
2. What is the degree of (A)? 40. 13m + n2 1m - 3n2 - 1m + 3n2 13m - n2
3. Add (B) and (C). 41. 1u + v2 3
4. Add (A) and (B). 42. 1x - y2 3
5. Subtract (B) from (C). 43. 1x - 2y2 3
6. Subtract (A) from (B). 44. 12m - n2 3
7. Multiply (B) and (C). 45. Subtract the sum of the last two polynomials from the sum
of the first two: 2x2 - 4xy + y 2, 3xy - y 2, x2 - 2xy - y 2,
8. Multiply (A) and (C).
- x2 + 3xy - 2y 2
In Problems 9–30, perform the indicated operations and simplify. 46. Subtract the sum of the first two polynomials from the sum
9. 21u - 12 - 13u + 22 - 212u - 32 of the last two: 3m2 - 2m + 5, 4m2 - m, 3m2 - 3m - 2,
m3 + m2 + 2
10. 21x - 12 + 312x - 32 - 14x - 52
11. 4a - 2a35 - 31a + 22 4 C In Problems 47–50, perform the indicated operations and simplify.
12. 2y - 3y34 - 21y - 12 4 47. 312x - 12 2 - x13x + 12 4 2

13. 1a + b2 1a - b2 48. 35x13x + 12 - 512x - 12 2 4 2
14. 1m - n2 1m + n2 49. 251x - 32 1x2 - 2x + 12 - x33 - x1x - 22 46
15. 13x - 52 12x + 12 50. - 3x5x3x - x12 - x2 4 - 1x + 22 1x2 - 32 6
16. 14t - 32 1t - 22 51. If you are given two polynomials, one of degree m and the
other of degree n, where m is greater than n, what is the
17. 12x - 3y2 1x + 2y2
degree of their product?
18. 13x + 2y2 1x - 3y2
52. What is the degree of the sum of the two polynomials in
19. 13y + 22 13y - 22 Problem 51?
20. 12m - 72 12m + 72 53. How does the answer to Problem 51 change if the two poly-
21. - 12x - 32 2 nomials can have the same degree?
22. - 15 - 3x2 2 54. How does the answer to Problem 52 change if the two poly-
nomials can have the same degree?
23. 14m + 3n2 14m - 3n2
55. Show by example that, in general, 1a + b2 2 ∙ a2 + b2.
24. 13x - 2y2 13x + 2y2 Discuss possible conditions on a and b that would make this
25. 13u + 4v2 2 a valid equation.
26. 14x - y2 2 56. Show by example that, in general, 1a - b2 2 ∙ a2 - b2.
Discuss possible conditions on a and b that would make this
27. 1a - b2 1a2 + ab + b2 2 a valid equation.
28. 1a + b2 1a2 - ab + b2 2
29. 31x - y2 + 3z431x - y2 - 3z4
30. 3a - 12b - c2 43a + 12b - c2 4
Applications
57. Investment. You have $10,000 to invest, part at 9% and the
B In Problems 31–44, perform the indicated operations and simplify.
rest at 12%. If x is the amount invested at 9%, write an alge-
31. m - 5m - 3m - 1m - 12 46 braic expression that represents the total annual income from
32. 2x - 35x + 23x - 1x + 52 4 + 16 both investments. Simplify the expression.
33. 1x2 - 2xy + y 2 2 1x2 + 2xy + y 2 2 58. Investment. A person has $100,000 to invest. If $x are
invested in a money market account yielding 7% and twice
34. 13x - 2y2 2 12x + 5y2 that amount in certificates of deposit yielding 9%, and if the
35. 15a - 2b2 2 - 12b + 5a2 2 rest is invested in high-grade bonds yielding 11%, write an
algebraic expression that represents the total annual income
36. 12x - 12 2 - 13x + 22 13x - 22 from all three investments. Simplify the expression.
SECTION B.3 Factoring Polynomials 579
59. Gross receipts. Four thousand tickets are to be sold for a 62. Nutrition. Each ounce of food M contains 8 units of calcium,
musical show. If x tickets are to be sold for $20 each and and each ounce of food N contains 5 units of calcium. A
three times that number for $30 each, and if the rest are sold 160-ounce diet mix is formed using foods M and N. If x is the
for $50 each, write an algebraic expression that represents the number of ounces of food M used, write an algebraic expres-
gross receipts from ticket sales, assuming all tickets are sold. sion that represents the total number of units of calcium in
Simplify the expression. the diet mix. Simplify the expression.
60. Gross receipts. Six thousand tickets are to be sold for a
concert, some for $20 each and the rest for $35 each. If x is Answers to Matched Problems
the number of $20 tickets sold, write an algebraic expression 1. (A) 5, 3, 0, 5 (B) 6, 4, 6
that represents the gross receipts from ticket sales, assuming 2. (A) 4u2 - v2 (B) - m3 - 3m2 + 2m - 4
all tickets are sold. Simplify the expression. (C) - x3 + 3x + 2
3. 3x4 - x3 - 5x2 + 2x - 2
61. Nutrition. Food mix A contains 2% fat, and food mix B 4. 3x2 + 5x - 10 5. 4x3 - 13x + 6
contains 6% fat. A 10-kilogram diet mix of foods A and B is 6. (A) 8u - 2uv - 3v (B) 4x2y 2 - 9 (C) m2 - 16n2
2 2
formed. If x kilograms of food A are used, write an algebraic (D) 4u2 - 12uv + 9v2 (E) 8x3 - 12x2y + 6xy 2 - y 3
expression that represents the total number of kilograms of 7. (A) - 2t 2 - 4t - 7 (B) - 3u2 - 6uv + 10v2
fat in the final food mix. Simplify the expression.
B.3 Factoring Polynomials

■■ Common Factors A positive integer is written in factored form if it is written as the product of two
or more positive integers; for example, 120 = 10 # 12. A positive integer is factored
completely if each factor is prime; for example, 120 = 2 # 2 # 2 # 3 # 5. (Recall that an
■■ Factoring by Grouping
■■ Factoring Second-Degree integer p 7 1 is prime if p cannot be factored as the product of two smaller positive
Polynomials integers. So the first ten primes are 2, 3, 5, 7, 11, 13, 17, 19, 23, and 29). A tree dia-
■■ Special Factoring Formulas gram is a helpful way to visualize a factorization (Fig 1).
■■ Combined Factoring Techniques 2
10
5
120
2
12 2
3
Figure 1
A polynomial is written in factored form if it is written as the product of two or

more polynomials. The following polynomials are written in factored form:
4x2y - 6xy 2 = 2xy12x - 3y2 2x3 - 8x = 2x1x - 221x + 22

x2 - x - 6 = 1x - 321x + 22 5m2 + 20 = 51m2 + 42
Unless stated to the contrary, we will limit our discussion of factoring

polynomials to polynomials with integer coefficients.
A polynomial with integer coefficients is said to be factored completely if each
factor cannot be expressed as the product of two or more polynomials with integer
coefficients, other than itself or 1. All the polynomials above, as we will see by the
conclusion of this section, are factored completely.
Writing polynomials in completely factored form is often a difficult task. But
accomplishing it can lead to the simplification of certain algebraic expressions and to
the solution of certain types of equations and inequalities. The distributive properties
for real numbers are central to the factoring process.
Common Factors
Generally, a first step in any factoring procedure is to factor out all factors common
to all terms.
EXAMPLE 1 Common Factors Factor out all factors common to all terms.
(A) 3x3y - 6x2y 2 - 3xy 3
(B) 3y12y + 52 + 212y + 52
SOLUTION
(A) 3x3y - 6x2y 2 - 3xy 3 = 13xy2x2 - 13xy22xy - 13xy2y 2
= 3xy1x2 - 2xy - y 2 2
(B) 3y12y + 52 + 212y + 52 = 3y12y ∙ 52 + 212y ∙ 52
= 13y + 2212y + 52
MATCHED PROBLEM 1 Factor out all factors common to all terms.
3 2 2 3
(A) 2x y - 8x y - 6xy (B) 2x13x - 22 - 713x - 22
Factoring by Grouping
Occasionally, polynomials can be factored by grouping terms in such a way that
we obtain results that look like Example 1B. We can then complete the factoring
following the steps used in that example. This process will prove useful in the next
subsection, where an efficient method is developed for factoring a second-degree
polynomial as the product of two first-degree polynomials, if such factors exist.
EXAMPLE 2 Factoring by Grouping Factor by grouping.

(A) 3x2 - 3x - x + 1
(B) 4x2 - 2xy - 6xy + 3y 2
(C) y 2 + xz + xy + yz
SOLUTION
(A) 3x2 - 3x - x + 1 Group the first two and the last two terms.
= 13x2 - 3x2 - 1x - 12 Factor out any common factors from each
= 3x1x ∙ 12 - 1x ∙ 12 group. The common factor 1x - 12 can be
= 1x ∙ 1213x - 12 taken out, and the factoring is complete.
(B) 4x2 - 2xy - 6xy + 3y 2 = 14x2 - 2xy2 - 16xy - 3y 2 2

= 2x12x ∙ y2 - 3y12x ∙ y2
= 12x ∙ y212x - 3y2
(C) If, as in parts (A) and (B), we group the first two terms and the last two terms
of y 2 + xz + xy + yz, no common factor can be taken out of each group to
complete the factoring. However, if the two middle terms are reversed, we can
proceed as before:
y 2 + xz + xy + yz = y 2 + xy + xz + yz
= 1y 2 + xy2 + 1xz + yz2
= y1y + x2 + z1x + y2
= y1x ∙ y2 + z1x ∙ y2
= 1x ∙ y21y + z2
MATCHED PROBLEM 2 Factor by grouping.

2
(A) 6x + 2x + 9x + 3
(B) 2u2 + 6uv - 3uv - 9v2
(C) ac + bd + bc + ad
Factoring Second-Degree Polynomials

We now turn our attention to factoring second-degree polynomials of the form
2x2 - 5x - 3 and 2x2 + 3xy - 2y 2
into the product of two first-degree polynomials with integer coefficients. Since many
second-degree polynomials with integer coefficients cannot be factored in this way,
it would be useful to know ahead of time that the factors we are seeking actually ex-
ist. The factoring approach we use, involving the ac test, determines at the beginning
whether first-degree factors with integer coefficients do exist. Then, if they exist, the
test provides a simple method for finding them.
THEOREM 1 ac Test for Factorability

If in polynomials of the form
ax2 + bx + c or ax2 + bxy + cy 2 (1)
the product ac has two integer factors p and q whose sum is the coefficient b of the
middle term; that is, if integers p and q exist so that
pq = ac and p + q = b (2)
then the polynomials have first-degree factors with integer coefficients. If no inte-
gers p and q exist that satisfy equations (2), then the polynomials in equations (1)
will not have first-degree factors with integer coefficients.
If integers p and q exist that satisfy equations (2) in the ac test, the factoring
always can be completed as follows: Using b = p + q, split the middle terms in
equations (1) to obtain
ax2 + bx + c = ax2 + px + qx + c
ax2 + bxy + cy 2 = ax2 + pxy + qxy + cy 2
Complete the factoring by grouping the first two terms and the last two terms as in
Example 2. This process always works, and it does not matter if the two middle terms
on the right are interchanged.
Several examples should make the process clear. After a little practice, you will
perform many of the steps mentally and will find the process fast and efficient.
EXAMPLE 3 Factoring Second-Degree Polynomials Factor, if possible, using integer

coefficients.
(A) 4x2 - 4x - 3 (B) 2x2 - 3x - 4 (C) 6x2 - 25xy + 4y 2
SOLUTION
(A) 4x2 - 4x - 3
Step 1 Use the ac test to test for factorability. Comparing 4x 2 - 4x - 3 with
ax2 + bx + c, we see that a = 4, b = - 4, and c = - 3. Multiply a and
c to obtain
pq
All factor pairs of ac = 1421 - 32 = - 12
112 1 -122
1 -12 1122
- 12 = ac List all pairs of integers whose product is - 12, as shown in the margin.
These are called factor pairs of - 12. Then try to find a factor pair that
122 1 -62
sums to b = - 4, the coefficient of the middle term in 4x2 - 4x - 3. (In
1 -22 162 practice, this part of Step 1 is often done mentally and can be done rather
132 1 -42 quickly.) Notice that the factor pair 2 and - 6 sums to - 4. By the ac test,
1 -32 142 4x2 - 4x - 3 has first-degree factors with integer coefficients.
Step 2 Split the middle term, using b = p + q, and complete the factoring by group-
ing. Using - 4 = 2 + 1 - 62, we split the middle term in 4x2 - 4x - 3 and
complete the factoring by grouping:
4x2 - 4x - 3 = 4x2 + 2x - 6x - 3
= 14x2 + 2x2 - 16x + 32
= 2x12x ∙ 12 - 312x ∙ 12
= 12x ∙ 1212x - 32
The result can be checked by multiplying the two factors to obtain the origi-
nal polynomial.
2
(B) 2x - 3x - 4
Step 1 Use the ac test to test for factorability:
ac = 1221 - 42 = - 8
pq
All factor pairs of Does - 8 have a factor pair whose sum is - 3? None of the factor pairs
1 -12 182
- 8 = ac listed in the margin sums to - 3 = b, the coefficient of the middle term in
112 1 -82 2x2 - 3x - 4. According to the ac test, we can conclude that 2x2 - 3x - 4
1 -22 142 does not have first-degree factors with integer coefficients, and we say that
122 1 -42 the polynomial is not factorable.
(C) 6x2 - 25xy + 4y 2
Step 1 Use the ac test to test for factorability:
ac = 162142 = 24
Mentally checking through the factor pairs of 24, keeping in mind that their
sum must be - 25 = b, we see that if p = - 1 and q = - 24, then
pq = 1 - 121 - 242 = 24 = ac
and
p + q = 1 - 12 + 1 - 242 = - 25 = b
So the polynomial is factorable.
Step 2 Split the middle term, using b = p + q, and complete the factoring by
grouping. Using - 25 = 1 - 12 + 1 - 242, we split the middle term in
6x2 - 25xy + 4y 2 and complete the factoring by grouping:
6x2 - 25xy + 4y 2 = 6x2 - xy - 24xy + 4y 2
= 16x2 - xy2 - 124xy - 4y 2 2
= x16x - y2 - 4y16x - y2
= 16x - y21x - 4y2
The check is left to the reader.
MATCHED PROBLEM 3 Factor, if possible, using integer coefficients.
2
(A) 2x + 11x - 6
(B) 4x2 + 11x - 6
(C) 6x2 + 5xy - 4y 2
Special Factoring Formulas

The factoring formulas listed in the following box will enable us to factor certain
polynomial forms that occur frequently. These formulas can be established by multi-
plying the factors on the right.
THEOREM 2 Special Factoring Formulas

Perfect square: 1. u2 + 2uv + v2 = 1u + v2 2
Perfect square: 2. u2 - 2uv + v2 = 1u - v2 2
Difference of squares: 3. u2 - v2 = 1u - v2 1u + v2
Difference of cubes: 4. u3 - v3 = 1u - v2 1u2 + uv + v2 2
Sum of cubes: 5. u3 + v3 = 1u + v2 1u2 - uv + v2 2
Notice that u2 + v2 is not included in the list of special factoring

! CAUTION
formulas. In fact,
u2 + v2 ∙ 1au + bv21cu + dv2
for any choice of real number coefficients a, b, c, and d.
EXAMPLE 4 Factoring Factor completely.

(A) 4m2 - 12mn + 9n2 (B) x2 - 16y 2 (C) z3 - 1
(D) m3 + n3 (E) a2 - 41b + 22 2
SOLUTION
(A) 4m2 - 12mn + 9n2 = 12m - 3n2 2
(B) x2 - 16y 2 = x2 - 14y2 2 = 1x - 4y21x + 4y2
(C) z3 - 1 = 1z - 121z2 + z + 12 Use the ac test to verify that z2 + z + 1
cannot be factored.
3 3 2 2
(D) m + n = 1m + n21m - mn + n 2 Use the ac test to verify that
m2 - mn + n2 cannot be factored.
(E) a2 - 41b + 22 2 = 3a - 21b + 2243a + 21b + 224
MATCHED PROBLEM 4 Factor completely:

2 2
(A) x + 6xy + 9y (B) 9x2 - 4y 2 (C) 8m3 - 1
(D) x3 + y 3z3 (E) 91m - 32 2 - 4n2
Combined Factoring Techniques

We complete this section by considering several factoring problems that involve
combinations of the preceding techniques.
PROCEDURE Factoring Polynomials

Step 1 Take out any factors common to all terms.
Step 2 Use any of the special formulas listed in Theorem 2 that are applicable.
Step 3 Apply the ac test to any remaining second-degree polynomial factors.
Note: It may be necessary to perform some of these steps more than once. Furthermore,
the order of applying these steps can vary.
EXAMPLE 5 Combined Factoring Techniques Factor completely.

(A) 3x3 - 48x (B) 3u4 - 3u3v - 9u2v2
(C) 3m2 - 24mn3 (D) 3x4 - 5x2 + 2
SOLUTION
(A) 3x3 - 48x = 3x1x2 - 162 = 3x1x - 421x + 42
(B) 3u4 - 3u3v - 9u2v2 = 3u2 1u2 - uv - 3v2 2
(C) 3m4 - 24mn3 = 3m1m3 - 8n3 2 = 3m1m - 2n21m2 + 2mn + 4n2 2

(D) 3x4 - 5x2 + 2 = 13x2 - 221x2 - 12 = 13x2 - 221x - 121x + 12
MATCHED PROBLEM 5 Factor completely.

3
(A) 18x - 8x
(B) 4m3n - 2m2n2 + 2mn3
(C) 2t 4 - 16t
(D) 2y 4 - 5y 2 - 12
Exercises B.3
A In Problems 1–8, factor out all factors common to all terms. 49. 5u2 + 4uv - 2v2 50. 3x2 - 2xy - 4y 2
1. 6m4 - 9m3 - 3m2 2. 6x4 - 8x3 - 2x2 51. 61x - y2 2 + 231x - y2 - 4
3 2 2 3 3 2 2 3
3. 8u v - 6u v + 4uv 4. 10x y + 20x y - 15xy 52. 41A + B2 2 - 51A + B2 - 6
5. 7m12m - 32 + 512m - 32 53. y 4 - 3y 2 - 4
6. 5x1x + 12 - 31x + 12 54. m4 - n4
7. 4ab12c + d2 - 12c + d2 55. 15y1x - y2 3 + 12x1x - y2 2
8. 12a1b - 2c2 - 15b1b - 2c2 56. 15x2 13x - 12 4 + 60x3 13x - 12 3
In Problems 9–18, factor by grouping. In Problems 57–60, discuss the validity of each statement. If the
2
9. 2x - x + 4x - 2 2
10. x - 3x + 2x - 6 statement is true, explain why. If not, give a counterexample.
2
11. 3y - 3y + 2y - 2 12. 2x2 - x + 6x - 3 57. If n is a positive integer greater than 1, then un - vn can be
factored.
13. 2x2 + 8x - x - 4 14. 6x2 + 9x - 2x - 3
58. If m and n are positive integers and m ∙ n, then um - vn is
15. wy - wz + xy - xz 16. ac + ad + bc + bd not factorable.
17. am - 3bm + 2na - 6bn 18. ab + 6 + 2a + 3b 59. If n is a positive integer greater than 1, then un + vn can be
factored.
B In Problems 19–56, factor completely. If a polynomial cannot be
factored, say so. 60. If k is a positive integer, then u2k + 1 + v2k + 1 can be factored.
19. 3y 2 - y - 2 20. 2x2 + 5x - 3
21. u2 - 2uv - 15v2 22. x2 - 4xy - 12y 2
23. m2 - 6m - 3 24. x2 + x - 4
25. w2x2 - y 2 26. 25m2 - 16n2 1. (A) 2xy1x2 - 4xy - 3y 2 2 (B) 12x - 72 13x - 22
2. (A) 13x + 12 12x + 32 (B) 1u + 3v2 12u - 3v2
27. 9m2 - 6mn + n2 28. x2 + 10xy + 25y 2
(C) 1a + b2 1c + d2
29. y 2 + 16 30. u2 + 81
3. (A) 12x - 12 1x + 62 (B) Not factorable
31. 4z2 - 28z + 48 32. 6x2 + 48x + 72 (C) 13x + 4y2 12x - y2
33. 2x4 - 24x3 + 40x2 34. 2y 3 - 22y 2 + 48y 4. (A) 1x + 3y2 2 (B) 13x - 2y2 13x + 2y2
2 2 2
35. 4xy - 12xy + 9x 36. 16x y - 8xy + y (C) 12m - 12 14m + 2m + 12
2
37. 6m - mn - 12n 2 2
38. 6s + 7st - 3t 2 (D) 1x + yz2 1x2 - xyz + y 2z2 2
(E) 331m - 32 - 2n4331m - 32 + 2n4
39. 4u3v - uv3 40. x3y - 9xy 3
5. (A) 2x13x - 22 13x + 22 (B) 2mn12m2 - mn + n2 2
41. 2x3 - 2x2 + 8x 42. 3m3 - 6m2 + 15m 2
(C) 2t1t - 22 1t + 2t + 42
43. 8x3 - 27y 3 44. 5x3 + 40y 3 (D) 12y 2 + 32 1y - 22 1y + 22
4 3
45. x y + 8xy 46. 8a - 1
C 47. 1x + 22 - 9y 2 2
48. 1a - b2 2 - 41c - d2 2
SECTION B.4 Operations on Rational Expressions 585
B.4 Operations on Rational Expressions

■■ Reducing to Lowest Terms We now turn our attention to fractional forms. A quotient of two algebraic expres-
■■ Multiplication and Division
sions (division by 0 excluded) is called a fractional expression. If both the numerator
and the denominator are polynomials, the fractional expression is called a rational
■■ Addition and Subtraction expression. Some examples of rational expressions are
■■ Compound Fractions 1 5 x + 7 x2 - 2x + 4
x3 + 2x x 3x2 - 5x + 1 1
In this section, we discuss basic operations on rational expressions. Since variables
represent real numbers in the rational expressions we will consider, the properties of
real number fractions summarized in Appendix B.1 will play a central role.
AGREEMENT Variable Restriction

Even though not always explicitly stated, we always assume that variables are
restricted so that division by 0 is excluded.
For example, given the rational expression

2x + 5
x1x + 221x - 32
the variable x is understood to be restricted from being 0, - 2, or 3, since these values
would cause the denominator to be 0.
Reducing to Lowest Terms
Central to the process of reducing rational expressions to lowest terms is the funda-
mental property of fractions, which we restate here for convenient reference:
THEOREM 1 Fundamental Property of Fractions

If a, b, and k are real numbers with b, k ∙ 0, then
ka a 5#2 2 x1x + 42 x
= = # x ∙ -4
b 5#7 7
=
kb 21x + 42 2
Using this property from left to right to eliminate all common factors from the
numerator and the denominator of a given fraction is referred to as reducing a frac-
tion to lowest terms. We are actually dividing the numerator and denominator by the
same nonzero common factor.
Using the property from right to left—that is, multiplying the numerator and de-
nominator by the same nonzero factor—is referred to as raising a fraction to higher
terms. We will use the property in both directions in the material that follows.
EXAMPLE 1 Reducing to Lowest Terms Reduce each fraction to lowest terms.

1#2#3#4 1#2#3#4 1 1
(A) # # # #
1 2 3 4 5 6# = # # # #
1 2 3 4 5 6# = # =
5 6 30
2#4#6#8 2#4#6#8
2 2 2 2
(B) # # # = # # # = 2 # 2 # 2 # 2 = 16
1 2 3 4 1 2 3 4
MATCHED PROBLEM 1 Reduce each fraction to lowest terms.

1#2#3#4#5 1 # 4 # 9 # 16
1#2#1#2#3 1#2#3#4
(A) (B)
CONCEPTUAL INSIGHT
Using Theorem 1 to divide the numerator and denominator of a fraction by a common
factor is often referred to as canceling. This operation can be denoted by drawing a
slanted line through each common factor and writing any remaining factors above or
below the common factor. Canceling is often incorrectly applied to individual terms in
the numerator or denominator, instead of to common factors. For example,
14 - 5 9 Theorem 1 does not apply. There are no common
=
2 2 factors in the numerator.
7 Incorrect use of Theorem 1. To cancel 2
14 - 5 14 - 5
∙ = 2 in the denominator, 2 must be a factor of
2 2 each term in the numerator.
1
EXAMPLE 2 Reducing to Lowest Terms Reduce each rational expression to lowest terms.
6x2 + x - 1 12x + 1213x - 12 Factor numerator and denominator
(A) 2 = completely.
2x - x - 1 12x + 121x - 12
3x - 1 Divide numerator and denominator by
=
x - 1 the common factor 12x + 12.
x4 - 8x x1x - 221x2 + 2x + 42
(B) =
3x3 - 2x2 - 8x x1x - 2213x + 42
x2 + 2x + 4
=
3x + 4
MATCHED PROBLEM 2 Reduce each rational expression to lowest terms.
x2 - 6x + 9 x3 - 1
(A) (B)
x2 - 9 x2 - 1
Multiplication and Division

Since we are restricting variable replacements to real numbers, multiplication and
division of rational expressions follow the rules for multiplying and dividing real
number fractions summarized in Appendix B.1.
THEOREM 2 Multiplication and Division

If a, b, c, and d are real numbers, then
a#c ac 3# x 3x
1. = , b, d ∙ 0 =
b d bd 5 x + 5 51x + 52
a c a#d
= #
3 x 3 x + 5
2. , = , b, c, d ∙ 0 ,
b d b c 5 x + 5 5 x
EXAMPLE 3 Multiplication and Division Perform the indicated operations and reduce to
lowest terms.
Factor numerators and denominators. Then divide
10x3y # x2 - 9
(A) any numerator and any denominator with a like
3xy + 9y 4x2 - 12x common factor.
5x2 1#1
10x3y # 1x - 321x + 32
=
3y1x + 32 4x1x - 32
3#1 2#1
2
5x
=
6
1
2 12 - x2
(B)
4 - 2x
, 1x - 22 = # 1 x - 2 =
x - 2
4 4 x - 2 1
2
-1
2 - x - 1x - 22
= = b - a = - 1a - b2, a useful
21x - 22 21x - 22 change in some problems
1
1
= -
2
MATCHED PROBLEM 3 Perform the indicated operations and reduce to lowest terms.
12x2y 3 # y 2 + 6y + 9 x2 - 16
(A) (B) 14 - x2 ,
2xy 2 + 6xy 3y 3 + 9y 2 5

Again, because we are restricting variable replacements to real numbers, addition and
subtraction of rational expressions follow the rules for adding and subtracting real
number fractions.
THEOREM 3 Addition and Subtraction

For a, b, and c real numbers,
a c a + c x 8 x + 8
1. + = , b ∙ 0 + =
b b b x + 5 x + 5 x + 5
a c a - c x x + 7 x - 1x + 72
2. - = , b ∙ 0 2 2
- 2 2
=
b b b 3x y 3x y 3x2y 2
We add rational expressions with the same denominators by adding or subtract-

ing their numerators and placing the result over the common denominator. If the
denominators are not the same, we raise the fractions to higher terms, using the fun-
damental property of fractions to obtain common denominators, and then proceed as
described.
Even though any common denominator will do, our work will be simplified if the
least common denominator (LCD) is used. Often, the LCD is obvious, but if it is not,
the steps in the next box describe how to find it.
PROCEDURE Least Common Denominator

The least common denominator (LCD) of two or more rational expressions is found
as follows:
1. Factor each denominator completely, including integer factors.
2. Identify each different factor from all the denominators.
3. Form a product using each different factor to the highest power that occurs in any
one denominator. This product is the LCD.
EXAMPLE 4 Addition and Subtraction Combine into a single fraction and reduce to lowest
terms.
3 5 11 4 5x 1 1 2
(A) + - (B) - 2 + 1 (C) - - 2
10 6 45 9x 6y x - 1 x x - 1
SOLUTION
(A) To find the LCD, factor each denominator completely:
10 = 2 # 5
6 = 2#3 s LCD = 2 # 32 # 5 = 90
45 = 32 # 5
Now use the fundamental property of fractions to make each denominator 90:
3 5 11 9#3 15 # 5 2 # 11
= #
15 # 6 2 # 45
+ - + -
10 6 45 9 10
27 75 22
= + -
90 90 90
27 + 75 - 22 80 8
= = =
90 90 9
9x = 32x
LCD = 2 # 32xy 2 = 18xy 2
6y 2 = 2 # 3y 2
(B) f
4 5x 2y2 # 4 3x # 5x 18xy2
- 2 + 1 = 2 - +
9x 6y 2y # 9x 3x # 6y 2 18xy2
8y 2 - 15x2 + 18xy 2
=
18xy 2
1 1 2
(C) - - 2
x - 1 x x - 1
1 1 2
= - - LCD = x1x - 12 1x + 12
x - 1 x 1x - 121x + 12
x1x + 12 - 1x - 121x + 12 - 2x
=
x1x - 121x + 12
x2 + x - x2 + 1 - 2x
=
x1x - 121x + 12
1 - x
=
x1x - 121x + 12
-1
- 1x - 12 -1
= =
x1x - 121x + 12 x1x + 12
1
MATCHED PROBLEM 4 Combine into a single fraction and reduce to lowest terms.
5 1 6 1 2x + 1 3
(A) - + (B) 2 - 3 +
28 10 35 4x 3x 12x
2 1 1
(C) 2 + -
x - 4x + 4 x x - 2
Compound Fractions
A fractional expression with fractions in its numerator, denominator, or both is called
a compound fraction. It is often necessary to represent a compound fraction as a
simple fraction—that is (in all cases we will consider), as the quotient of two poly-
nomials. The process does not involve any new concepts. It is a matter of applying
old concepts and processes in the correct sequence.
EXAMPLE 5 Simplifying Compound Fractions Express as a simple fraction reduced to lowest

terms:
1 1 y x
- - 2
5 + h 5 x2 y
(A) (B)
h y x
-
x y
SOLUTION We will simplify the expressions in parts (A) and (B) using two differ-
ent methods—each is suited to the particular type of problem.
(A) We simplify this expression by combining the numerator into a single fraction
and using division of rational forms.
1 1
-
5 + h 5 1 1 h
= c - d ,
h 5 + h 5 1
5 - 5 - h 1
= ~
515 + h2 h
-h -1
= =
515 + h2h 515 + h2
(B) The method used here makes effective use of the fundamental property of
fractions in the form
a ka
= b, k ∙ 0
b kb
Multiply the numerator and denominator by the LCD of all fractions in the
numerator and denominator—in this case, x2 y 2:
y x y x
x 2y 2 a 2 - 2b x 2y 2 2 - x 2y 2 2
x y x y y 3 - x3
= = 3
y x y x xy - x3y
x 2y 2 a - b x 2y 2 - x 2y 2
x y x y
1
1y - x2 1y 2 + xy + x2 2
=
xy1y - x21y + x2
1
y 2 + xy + x2 x2 + xy + y 2
= or
xy1y + x2 xy1x + y2
MATCHED PROBLEM 5 Express as a simple fraction reduced to lowest terms:

1 1 a b
- -
2 + h 2 b a
(A) (B)
h a b
+ 2 +
b a
Exercises B.4
A In Problems 1–22, perform the indicated operations and reduce 12 # 11 # 10 # 9 15 # 10 # 5
4#3#2#1 20 # 15 # 10
3. 4.
answers to lowest terms.
5 # 9 # 13 10 # 9 # 8 d5 d2 a d5 d2 # a
1.
3#5#7
2.
3#2#1 5. , a 2 # 3b 6. a , b
3a 6a 4d 3a 6a2 4d 3
x2 x 1 2y -1 y In Problems 35–42, imagine that the indicated “solutions” were

7. + - 8. - -
12 18 30 18 28 42 given to you by a student whom you were tutoring in this class.
4m - 3 3 2m - 1 (A) Is the solution correct? If the solution is incorrect, explain
9. 3
+ - what is wrong and how it can be corrected.
18m 4m 6m2
3x + 8 2x - 1 5 (B) Show a correct solution for each incorrect solution.
10. 2
- 3
-
4x x 8x
x2 + 4x + 3 x2 + 4x
x2 - 9 35. = = x + 4
11. , 1x2 - x - 122 x + 3 x
x2 - 3x
2x2 + 7x + 3 x2 - 3x - 4 x2 - 3x
12. , 1x + 32 36. = = x - 3
4x2 - 1 x - 4 x
2 1 5 3 1x + h2 2 - x2
13. - 14. - 37. = 1x + 12 2 - x2 = 2x + 1
x x - 3 m - 2 2m + 1 h
2 5 1x + h2 3 - x3
15. 2
- 2 38. = 1x + 12 3 - x3 = 3x2 + 3x + 1
1x + 12 x - x - 2 h
3 5
16. 2 - x2 - 3x x2 - 3x + x - 3
x - 5x + 6 1x - 22 2 39. 2
+ x - 3 = = 1
x - 2x - 3 x2 - 2x - 3
x + 1 m - 1
17. - 1 18. m - 3 - 2 x + 3 2x + 2 - x - 3 1
x - 1 m - 2 40. - 2 = =
x - 1 2 x + 1
3 2 5 2 x - 1 x - 1
19. - 20. -
a - 1 1 - a x - 3 3 - x 2x2 x 2x2 - x2 - 2x x
41. 2
- = 2
=
2x x - 4 m + 2 m x - 4 x - 2 x - 4 x + 2
21. 2
- 2 22. 2
- 2
x - 16 x + 4x m - 2m m - 4 x - 2 x + x - 2 2
42. x + = 2 =
B In Problems 23–34, perform the indicated operations and x2 - 3x + 2 x - 3x + 2 x - 2
reduce answers to lowest terms. Represent any compound
fractions as simple fractions reduced to lowest terms. C Represent the compound fractions in Problems 43–46 as simple
fractions reduced to lowest terms.
x2 x - 1 1 1 1
23. 2
+ - - 2
x + 2x + 1 3x + 3 6 31x + h2 2 3x
y 1 2 43.
24. - 2 - 2 h
y2 - y - 2 y + 5y - 14 y + 8y + 7 1 1
x - 2
1 - 1x + h2 2 x
y 2 44.
25. 26. h
y 3
2 - 5 - 2
x 4x + 1 45. x -
1
c + 2 c - 2 c 1 -
27. - + x
5c - 5 3c - 3 1 - c
1
x + 7 y + 9 46. 2 -
28. + 2
ax - bx by - ay 1 -
a + 2
3 y2
1 + 1 -
x x2 Answers to Matched Problems
29. 30.
9 y
x - 1 - 1. (A) 10 (B) 24
x x
1 1 1 1 x - 3 x2 + x + 1
- - 2. (A) (B)
21x + h2 2x x + h x x + 3 x + 1
31. 32. -5
h h 3. (A) 2x (B)
x y 2 15 x + 4
- 2 + 1 + - 1 3x2 - 5x - 4 4
y x x x2 4. (A) (B) (C)
33. 34. 4 12x3 x1x - 22 2
x y 4 5
- 1 + -
y x x x2 -1 a - b
5. (A) (B)
212 + h2 a + b
SECTION B.5 Integer Exponents and Scientific Notation 591
B.5 Integer Exponents and Scientific Notation

■■ Integer Exponents We now review basic operations on integer exponents and scientific notation.
■■ Scientific Notation
Integer Exponents
DEFINITION Integer Exponents

For n an integer and a a real number:
1. For n a positive integer,
an = a # a # g # a n factors of a 54 = 5 # 5 # 5 # 5
2. For n = 0,
a0 = 1 a ∙ 0 120 = 1
00 is not defined.
3. For n a negative integer,
1 1 1
an = a ∙ 0 a -3 = =
a -n a -1-32 a3
[If n is negative, then 1 - n2 is positive.]
Note: It can be shown that for all integers n,
1 1 1 1
a -n = n and an = -n a ∙ 0 a5 = a -5 =
a a a -5′ a5
The following properties are very useful in working with integer exponents.
THEOREM 1 Exponent Properties

For n and m integers and a and b real numbers,
1. aman = am + n a8a -3 = a8 + 1-32 = a5
n m mn
2. 1a 2 = a 1a -2 2 3 = a31-22 = a -6
m m m
3. 1ab2 = a b 1ab2 -2 = a -2b-2
m m
a a a 5 a5
4. a b = m b ∙ 0 a b = 5
b b b b
am 1 a -3
1 1
5. n = am - n = n - m a ∙ 0 = 7 - 1-32 = 10
a a a7 a a
Exponents are frequently encountered in algebraic applications. You should

sharpen your skills in using exponents by reviewing the preceding basic definitions
and properties and the examples that follow.
EXAMPLE 1 Simplifying Exponent Forms Simplify, and express the answers using positive
exponents only.
(A) 12x3 213x5 2 = 2 # 3x3 + 5 = 6x8
1
(B) x5x -9 = x -4 =
x4
x5 1 x5 1 1
(C) = x5 - 7 = x -2 = 2 or = 7-5 = 2
x7 x x7 x x
x -3 y4
(D) =
y -4 x3
u6
(E) 1u-3v2 2 -2 = 1u-3 2 -2 1v2 2 -2 = u6v -4 =
v4
y -5 -2 1y -5 2 -2 y 10
(F) a b = = = y6
y -2 1y -2 2 -2 y4
4m-3n-5 2m-3 - 1-42 2m
(G) -4 3 = 3 - 1-52
= 8
6m n 3n 3n
MATCHED PROBLEM 1 Simplify, and express the answers using positive expo-
nents only.
(A) 13y 4 212y 3 2 (B) m2m-6 (C) 1u3v -2 2 -2
y -6 -1 8x -2y -4
(D) a b (E)
y -2 6x -5y 2
1 - x
EXAMPLE 2 Converting to a Simple Fraction Write -1 as a simple fraction with posi-
tive exponents. x - 1
SOLUTION First note that

1 - x x11 - x2
-1 ∙ A common error
x - 1 -1
The original expression is a compound fraction, and we proceed to simplify it as
follows:
1 - x 1 - x
-1 = Multiply numerator and denominator
x - 1 1 by x to clear internal fractions.
- 1
x
x11 - x2
=
1
xa - 1b
x
x11 - x2
= = x
1 - x
1 + x -1
MATCHED PROBLEM 2 Write as a simple fraction with positive exponents.
1 - x -2
Scientific Notation
In the real world, one often encounters very large and very small numbers. For
example,
• The public debt in the United States in 2016, to the nearest billion dollars, was
$19,573,000,000,000
• The world population in the year 2025, to the nearest million, is projected to be
7,947,000,000
• The sound intensity of a normal conversation is
0.000 000 000 316 watt per square centimeter
SECTION B.5 Integer Exponents and Scientific Notation 593
It is generally troublesome to write and work with numbers of this type in stan-
dard decimal form. The first and last example cannot even be entered into many cal-
culators as they are written. But with exponents defined for all integers, we can now
express any finite decimal form as the product of a number between 1 and 10 and an
integer power of 10, that is, in the form
a * 10n 1 … a 6 10, a in decimal form, n an integer
A number expressed in this form is said to be in scientific notation. The follow-
ing are some examples of numbers in standard decimal notation and in scientific
notation:
Decimal and Scientific Notation

7 = 7 * 100 0.5 = 5 * 10-1
67 = 6.7 * 10 0.45 = 4.5 * 10-1
580 = 5.8 * 102 0.0032 = 3.2 * 10-3
43,000 = 4.3 * 104 0.000 045 = 4.5 * 10-5
73,400,000 = 7.34 * 107 0.000 000 391 = 3.91 * 10-7
Note that the power of 10 used corresponds to the number of places we move the
decimal to form a number between 1 and 10. The power is positive if the decimal is
moved to the left and negative if it is moved to the right. Positive exponents are as-
sociated with numbers greater than or equal to 10, negative exponents are associated
with positive numbers less than 1, and a zero exponent is associated with a number
that is 1 or greater but less than 10.
EXAMPLE 3 Scientific Notation

(A) Write each number in scientific notation:
7,320,000 and 0.000 000 54
(B) Write each number in standard decimal form:
4.32 * 106 and 4.32 * 10 -5
SOLUTION
(A) 7,320,000 = 7.320 000. * 106 = 7.32 * 106
6 places left
Positive exponent
0.000 000 54 = 0.000 000 5.4 * 10 -7 = 5.4 * 10 -7

7 places right
Negative exponent
4.32
(B) 4.32 * 106 = 4,320,000 4.32 * 10 -5 = = 0.000 043 2
6 places right 105
5 places left
Positive exponent 6 Negative exponent - 5
Matched Problem 3
(A) Write each number in scientific notation: 47,100; 2,443,000,000; 1.45
(B) Write each number in standard decimal form: 3.07 * 108; 5.98 * 10 -6
Exercises B.5
A In Problems 1–14, simplify and express answers using positive 45. 2x -2 1x - 12 - 2x -3 1x - 12 2
exponents only. Variables are restricted to avoid division by 0.
3 46. 2x1x + 32 -1 - x2 1x + 32 -2
1. 2x -9 2. 3y -5 3.
2w -7 In Problems 47–50, convert each number to scientific notation
5
4. 5. 2x -8x5 6. 3c -9c4 and simplify. Express the answer in both scientific notation and in
4x -9 standard decimal form.
w -8 m-11
7. 8. 9. 12a -3 2 2 9,600,000,000
w -3 m-5 47.
11,600,0002 10.000 000 252
10. 7d -4d 4 11. 1a -3 2 2 12. 15b-2 2 2
160,0002 10.000 0032
13. 12x4 2 -3 14. 1a -3b4 2 -3 48.
10.00042 11,500,0002
In Problems 15–20, write each number in scientific notation. 11,250,0002 10.000 382
49.
15. 82,300,000,000 16. 5,380,000 0.0152
17. 0.783 18. 0.019 10.000 000 822 1230,0002
50.
19. 0.000 034 20. 0.000 000 007 832 1625,0002 10.00822
2
In Problems 21–28, write each number in standard decimal 51. What is the result of entering 23 on a calculator?
notation. 2
52. Refer to Problem 51. What is the difference between 213 2 and
2
21. 4 * 104 22. 9 * 106 123 2 2? Which agrees with the value of 23 obtained with a
23. 7 * 10-3 24. 2 * 10-5 calculator?
25. 6.171 * 107 26. 3.044 * 103 53. If n = 0, then property 1 in Theorem 1 implies that
27. 8.08 * 10-4 28. 1.13 * 10-2 ama0 = am + 0 = am. Explain how this helps motivate the
definition of a0.
B In Problems 29–38, simplify and express answers using positive
54. If m = - n, then property 1 in Theorem 1 implies that
exponents only. Assume that variables are nonzero.
a -nan = a0 = 1. Explain how this helps motivate the defini-
29. 122 + 312 0 30. 12x3y 4 2 0 tion of a -n.
10-3 # 104 10-17 # 10-5
31. 32. C Write the fractions in Problems 55–58 as simple fractions
10-11 # 10-2 10-3 # 10-14 reduced to lowest terms.
33. 15x2y -3 2 -2 34. 12m-3n2 2 -3 u + v x -2 - y -2
-2 -3 55. 56.
-5 2a u + v -1
-1
x -1 + y -1
35. a b 36. a b
2x3 3b2 -2
b - c -2 xy -2 - yx -2
8x -3y -1 57. 58.
9m-4n3 b-3 - c -3 y -1 - x -1
37. 2 -4
38.
6x y 12m-1n-1
In Problems 39–42, write each expression in the form axp + bxq
or axp + bxq + cxr, where a, b, and c are real numbers and p, q, Applications
and r are integers. For example,
Problems 59 and 60 refer to Table 1.
2x4 - 3x2 + 1 2x4 3x2 1 3 1
= 3 - 3 + 3 = x - x -1 + x -3 Table 1 U.S. Public Debt, Interest on Debt, and Population
2x3 2x 2x 2x 2 2
Year Public Debt ($) Interest on Debt ($) Population
7x5 - x2 5x3 - 2
39. 40. 2000 5,674,000,000,000 362,000,000,000 281,000,000
4x5 3x2 2016 19,573,000,000,000 433,000,000,000 323,000,000
4 2
5x - 3x + 8 2x - 3x2 + x
3
41. 2
42. 59. Public debt. Carry out the following computations using scien-
2x 2x2
tific notation, and write final answers in standard decimal form.
Write each expression in Problems 43–46 with positive exponents
only, and as a single fraction reduced to lowest terms. (A) What was the per capita debt in 2016 (to the nearest
dollar)?
3x2 1x - 12 2 - 2x3 1x - 12
43. (B) What was the per capita interest paid on the debt in 2016
1x - 12 4 (to the nearest dollar)?
5x4 1x + 32 2 - 2x5 1x + 32 (C) What was the percentage interest paid on the debt in
44. 4 2016 (to two decimal places)?
1x + 32
SECTION B.6 Rational Exponents and Radicals 595
60. Public debt. Carry out the following computations using sci- 61. 9 ppm, the standard for carbon monoxide, when averaged
entific notation, and write final answers in standard decimal over a period of 8 hours
form. 62. 0.03 ppm, the standard for sulfur oxides, when averaged over
(A) What was the per capita debt in 2000 (to the nearest a year
dollar)? 63. Crime. In 2015, the United States had a violent crime rate of
(B) What was the per capita interest paid on the debt in 2000 373 per 100,000 people and a population of 320 million peo-
(to the nearest dollar)? ple. How many violent crimes occurred that year? Compute
the answer using scientific notation and convert the answer to
(C) What was the percentage interest paid on the debt in standard decimal form (to the nearest thousand).
2000 (to two decimal places)?
64. Population density. The United States had a 2016 popula-
Air pollution. Air quality standards establish maximum amounts
tion of 323 million people and a land area of 3,539,000
of pollutants considered acceptable in the air. The amounts are
square miles. What was the population density? Compute the
frequently given in parts per million (ppm). A standard of 30 ppm
answer using scientific notation and convert the answer to
also can be expressed as follows:
standard decimal form (to one decimal place).
30 3 * 10
30 ppm = = Answers to Matched Problems
1,000,000 106
1 v4
= 3 * 10 -5
= 0.000 03 = 0.003% 1. (A) 6y 7 (B) (C)
m4 u6
In Problems 61 and 62, express the given standard: 4x3
(D) y 4 (E)
(A) In scientific notation 3y 6
x
2.
(B) In standard decimal notation x - 1
(C) As a percent 3. (A) 4.7 * 104; 2.443 * 109; 1.45 * 100
(B) 307,000,000; 0.000 005 98
B.6 Rational Exponents and Radicals

■■ nth Roots of Real Numbers Square roots may now be generalized to nth roots, and the meaning of exponent may
■■ Rational Exponents and Radicals
be generalized to include all rational numbers.
■■ Properties of Radicals nth Roots of Real Numbers
Consider a square of side r with area 36 square inches. We can write
r 2 = 36
and conclude that side r is a number whose square is 36. We say that r is a square
root of b if r 2 = b. Similarly, we say that r is a cube root of b if r 3 = b. And, in
general,
DEFINITION nth Root

For any natural number n,
r is an nth root of b if r n = b
So 4 is a square root of 16, since 42 = 16; - 2 is a cube root of - 8, since

1 - 22 3 = - 8. Since 1 - 42 2 = 16, we see that - 4 is also a square root of 16. It can
be shown that any positive number has two real square roots; two real 4th roots; and,
in general, two real nth roots if n is even. Negative numbers have no real square roots;
no real 4th roots; and, in general, no real nth roots if n is even. The reason is that no
real number raised to an even power can be negative. For odd roots, the situation is
simpler. Every real number has exactly one real cube root; one real 5th root; and, in
general, one real nth root if n is odd.
Additional roots can be considered in the complex number system. In this book,
we restrict our interest to real roots of real numbers, and root will always be inter-
preted to mean “real root.”
Rational Exponents and Radicals

We now turn to the question of what symbols to use to represent nth roots. For n a
natural number greater than 1, we use
n
b1>n or 2b
to represent a real nth root of b. The exponent form is motivated by the fact that
1b1>n 2 n = b if exponent laws are to continue to hold for rational exponents. The
other form is called an nth root radical. In the expression below, the symbol 2 is
called a radical, n is the index of the radical, and b is the radicand:
Index Radical
n
2b
Radicand
When the index is 2, it is usually omitted. That is, when dealing with square roots,
2 n
we simply use 1b rather than 2 b. If there are two real nth roots, both b1>n and 2b
denote the positive root, called the principal nth root.
EXAMPLE 1 Finding nth Roots Evaluate each of the following:

(A) 41>2 and 24 (B) - 41>2 and - 24 (C) 1 - 42 1>2 and 1 - 4
3 3 3
(D) 81>3 and 2 8 (E) 1 - 82 1>3 and 2 -8 (F) - 81>3 and - 2 8
SOLUTION
(A) 41>2 = 14 = 2 1 14 ∙ { 22 (B) - 41>2 = - 14 = - 2
(C) 1 - 42 1>2 and 1 - 4 are not real numbers
3 3
(D) 81>3 = 2 8 = 2 (E) 1 - 82 1>3 = 2 -8 = -2
3
(F) - 81>3 = - 2 8 = -2
MATCHED PROBLEM 1 Evaluate each of the following:

(A) 161>2 (B) - 116 (C) 23
- 27 (D) 1 - 92 1>2 (E) 12
4
81 2 3
The symbol 24 represents the single number 2, not { 2. Do not

! CAUTION confuse 24 with the solutions of the equation x2 = 4, which are
usually written in the form x = {14 = { 2.
We now define br for any rational number r = m>n.
DEFINITION Rational Exponents

If m and n are natural numbers without common prime factors, b is a real number,
and b is nonnegative when n is even, then
m
m>n 1 b1>n 2 m = 1 2b 2
n
82>3 = 181>3 2 2 = (23
8)2 = 22 = 4
b = e m 1>n
1b 2 n
= 2bm 2>3
8 = 18 2 2 1>3 3 2
= 28 = 2 3
64 = 4
and
1 1 1
b - m>n = m>n
b ∙ 0 8 - 2>3 = 2>3
=
b 8 4
m>n
Note that the two definitions of b are equivalent under the indicated restrictions
on m, n, and b.
CONCEPTUAL INSIGHT
All the properties for integer exponents listed in Theorem 1 in Section B.5 also
hold for rational exponents, provided that b is nonnegative when n is even. This
restriction on b is necessary to avoid nonreal results. For example,
1 - 42 3>2 = 21 - 42 3 = 2- 64 Not a real number
To avoid nonreal results, all variables in the remainder of this discussion represent
positive real numbers.
EXAMPLE 2 From Rational Exponent Form to Radical Form and Vice Versa Change rational
exponent form to radical form.
7
(A) x1>7 = 2x
(B) 13u2v3 2 3>5 = 2 13u2v3 2 3 or 1 23u2v3 2 3 The first is usually preferred.
5 5
1 1 3 -2 3 1
(C) y - 2>3 = = or 2y or
y 2>3
2y
3 2 A y2
Change radical form to rational exponent form.
5 3
(D) 26 = 61>5 (E) - 2x2 = - x2>3
(F) 2x2 + y 2 = 1x2 + y 2 2 1>2 Note that 1x2 + y 2 2 1>2 ∙ x + y. Why?
MATCHED PROBLEM 2 Convert to radical form.
(A) u1>5 (B) 16x2y 5 2 2>9 (C) 13xy2 - 3>5
Convert to rational exponent form.
4 7 3 3
(D) 29u (E) - 2 12x2 4 (F) 2x + y3
EXAMPLE 3 Working with Rational Exponents Simplify each and express answers using
positive exponents only. If rational exponents appear in final answers, convert to
radical form.
6
(A) 13x1>3 212x1>2 2 = 6x1>3 + 1>2 = 6x5>6 = 6 2x5
(B) 1 - 82 5>3 = 31 - 82 1>3 4 5 = 1 - 22 5 = - 32
8x
(C) 1 2x1>3y - 2>3 2 3 = 8xy - 2 = 2
y
4x1>3 1>2 41>2x1>6 2 2 2
(D) a 1>2
b = 1>4
= 1>4 - 1>6 = 1>12 = 12
x x x x 2x
MATCHED PROBLEM 3 Simplify each and express answers using positive expo-
nents only. If rational exponents appear in final answers, convert to radical form.
(A) 93>2 (B) 1 - 272 4>3 (C) 1 5y 1>4 21 2y 1>3 2
8x1>2 1>3
(D) 1 2x - 3>4y 1>4 2 4 (E) a b
x2>3
EXAMPLE 4 Working with Rational Exponents Multiply, and express answers using positive
exponents only.
(A) 3y 2>3 1 2y 1>3 - y 2 2 (B) 1 2u1>2 + v1>2 21 u1>2 - 3v1>2 2
SOLUTION
(A) 3y 2>3 12y 1>3 - y 2 2 = 6y 2>3 + 1>3 - 3y 2>3 + 2
= 6y - 3y 8>3
(B) 1 2u1>2 + v1>2 21 u1>2 - 3v1>2 2 = 2u - 5u1>2v1>2 - 3v
MATCHED PROBLEM 4 Multiply, and express answers using positive exponents only.
1>4 3 3>4
(A) 2c 15c - c 2 (B) 17x1>2 - y 1>2 212x1>2 + 3y 1>2 2
EXAMPLE 5 Working with Rational Exponents Write the following expression in the form
ax p + bxq, where a and b are real numbers and p and q are rational numbers:
3 2
2 2x - 3 2x
3
22 x
3 2
2 2x - 3 2x 2x1>2 - 3x2>3
SOLUTION = Change to rational exponent form.
3
2 2x 2x1>3
2x1>2 3x2>3
= 1>3 - 1>3 Separate into two fractions.
2x 2x
= x1>6 - 1.5x1>3
MATCHED PROBLEM 5 Write the following expression in the form axp + bxq,
where a and b are real numbers and p and q are rational numbers:
3
52 x - 4 2x
2 2x3
Properties of Radicals
Changing or simplifying radical expressions is aided by several properties of radicals
that follow directly from the properties of exponents considered earlier.
THEOREM 1 Properties of Radicals

If n is a natural number greater than or equal to 2, and if x and y are positive real
numbers, then
n
1. 2x n = x 2 3 3
x = x
n n n 5 5 5
2. 1xy = 1x 1y 2 xy = 2 x 2 y
n
n x 1x 4 x
4
2x
3. = n = 4
7 y 1y A y 2y
EXAMPLE 6 Applying Properties of Radicals Simplify using properties of radicals.

4 4 4 3 xy
(A) 213x4y 3 2 4 (B) 28 22 (C)
A 27
SOLUTION
4
(A) 213x4y 3 2 4 = 3x4y 3 Property 1
4 4 4 4 4
(B) 2 82 2 = 2 16 = 22 = 2 Properties 2 and 1
3 3
3 xy 1xy 1xy 13
(C) = 3 = or 1xy Properties 3 and 1
A 27 127 3 3
MATCHED PROBLEM 6 Simplify using properties of radicals.

3
7 3 2 16x4y
(A) 2 1x3 + y 3 2 7 (B) 2 8y 3 (C) 3
2 2xy
What is the best form for a radical expression? There are many answers, depend-
ing on what use we wish to make of the expression. In deriving certain formulas, it
is sometimes useful to clear either a denominator or a numerator of radicals. The
process is referred to as rationalizing the denominator or numerator. Examples 7 and

8 illustrate the rationalizing process.
EXAMPLE 7 Rationalizing Denominators Rationalize each denominator.

6x 6 x - 4
(A) (B) (C)
12x 17 - 15 1x + 2
SOLUTION
6x 6x # 12x 6x 12x
(A) = = = 3 12x
12x 12x 12x 2x
(B)
6
=
6 # 17 + 15
17 - 15 17 - 15 17 + 15
61 17 + 152
= = 31 17 + 152
2
x - 4 x - 4 # 1x - 2
(C) =
1x + 2 1x + 2 1x - 2
1x - 42(1x - 2)
= = 1x - 2
x - 4
MATCHED PROBLEM 7 Rationalize each denominator.
12ab2 9 x2 - y 2
(A) (B) (C)
13ab 16 + 13 1x - 1y
EXAMPLE 8 Rationalizing Numerators Rationalize each numerator.

12 3 + 1m 12 + h - 12
(A) (B) (C)
2 13 9 - m h
SOLUTION
12 12 # 12 2 1
(A) = = =
2 13 2 13 12 2 16 16
3 + 1m 3 + 1m # 3 - 1m 9 - m 1
(B) = = =
9 - m 9 - m 3 - 1m 19 - m213 - 1m2 3 - 1m
12 + h - 12 12 + h - 12 # 12 + h + 12
(C) =
h h 12 + h + 12
h 1
= =
h1 12 + h + 122 12 + h + 12
MATCHED PROBLEM 8 Rationalize each numerator.

13 2 - 1n 13 + h - 13
(A) (B) (C)
3 12 4 - n h
Exercises B.6
A Change each expression in Problems 1–6 to radical form. Do not Change each expression in Problems 7–12 to rational exponent
simplify. form. Do not simplify.
4 3 5 2 5
1. 6x3>5 2. 7y 2>5 3. 132x2y 3 2 3>5 7. 52x 8. 7m2n 9. 2 12x2y2 3
7 3 3 3 2
4. 17x2y2 5>7 5. 1x2 + y 2 2 1>2 6. x1>2 + y 1>2 10. 2 18x4y2 3 11. 2 x + 2 y 12. 2x + y3
In Problems 13–24, find rational number representations for each, 31x + 12 71x - y2 2 3a - 3b
if they exist. 58. 59. 60.
1x + 4 1x - 1y 1a + 1b
13. 251>2 14. 641>3 15. 163>2
Rationalize the numerators in Problems 61–66.
16. 163>4 17. - 491>2 18. 1 - 492 1>2
15xy 13mn
4 3>2 61. 62.
19. - 642>3 20. 1 - 642 2>3 21. a b 2 2
5x y 3mn
25
8 2>3 1x + h - 1x 121a + h2 - 12a
22. a b 23. 9-3>2 24. 8-2>3 63. 64.
27 h h
In Problems 25–34, simplify each expression and write answers 1t - 1x 1x - 1y
65. 2 66.
using positive exponents only. All variables represent positive real t - x2 1x + 1y
numbers.
m2>3 Problems 67–70 illustrate common errors involving rational ex-
25. x4>5x -2>5 26. y -3>7y 4>7 27. -1>3
m ponents. In each case, find numerical examples that show that the
x1>4 left side is not always equal to the right side.
3 -6 1>3
28. 3>4 29. 18x y 2 30. 14u-2v4 2 1>2
x 67. 1x + y2 1>2 ∙ x1>2 + y 1>2 68. 1x3 + y 3 2 1>3 ∙ x + y
4x -2 -1>2 w4 -1>2 1 1
31. a 4 b 32. a -2 b 69. 1x + y2 1>3 ∙ 70. 1x + y2 -1>2 ∙
y 9x 1x + y2 3 1x + y2 2
18x2 -1>3 6a3>4
33. 34. C In Problems 71–82, discuss the validity of each statement. If the
12x1>4 15a -1>3
statement is true, explain why. If not, give a counterexample.
Simplify each expression in Problems 35–40 using properties of 71. 2x2 = x for all real numbers x
radicals. All variables represent positive real numbers.
72. 2x2 = ∙ x ∙for all real numbers x
5 3
35. 2 12x + 32 5 36. 2 17 + 2y2 3 3 3
73. 2x = ∙ x ∙for all real numbers x
3 7 5 4 5 2 5 3
37. 26x 215x 230x 38. 216a 24a 28a 3 3
74. 2x = x for all real numbers x
16x 110 18 112y
39. 40. 75. If r 6 0, then r has no cube roots.
115x 16y
76. If r 6 0, then r has no square roots.
B In Problems 41–48, multiply, and express answers using positive
exponents only. 77. If r 7 0, then r has two square roots.
41. 3x3>4 14x1>4 - 2x8 2 78. If r 7 0, then r has three cube roots.
42. 2m1>3 13m2>3 - m6 2 79. The fourth roots of 100 are 110 and - 110.
43. 13u1>2 - v1>2 2 1u1>2 - 4v1>2 2 80. The square roots of 216 - 5 are 13 - 12 and 12 - 13.
44. 1a1>2 + 2b1>2 2 1a1>2 - 3b1>2 2 81. 2355 - 60135 = 517 - 615
3
45. 16m1>2 + n - 1>2 2 16m - n - 1>2 2 82. 2 7 - 512 = 1 - 12
46. 12x - 3y 1>3 2 12x1>3 + 12 In Problems 83–88, simplify by writing each expression as a
simple or single fraction reduced to lowest terms and without
47. 13x1>2 - y 1>2 2 2
negative exponents.
48. 1x1>2 + 2y 1>2 2 2 1
83. - 1x - 22 1x + 32 -3>2 + 1x + 32-1>2
2
Write each expression in Problems 49–54 in the form axp + bxq,
where a and b are real numbers and p and q are rational numbers. 1
84. 21x - 22 -1>2 - 12x + 32 1x - 22-3>2
2
3 2 4 3 3
2x + 2 121x - 3 22x + 2 x
49. 50. 51. 1x - 12 1>2 - x1 12 2 1x - 12 -1>2
3
22 x 41x 3x 85.
x - 1
3 2 3 2
32x + 1x 22x - 1x x - 41x 12x - 12 1>2
- 1x + 22 1 12 2 12x - 12-1>2 122
52. 53. 54. 3 86.
5x 41x 21x 2x - 1
Rationalize the denominators in Problems 55–60. 1x + 22 2>3 - x 1 23 2 1x + 22 -1>3

87.
1x + 22 4>3
12mn2 14x2 21x + 32
55. 56. 57. 213x - 12 1>3 - 12x + 12 1 13 2 13x - 12 -2>3 132
13mn 17x 1x - 2 88.
13x - 12 2>3
SECTION B.7 Quadratic Equations 601
In Problems 89–94, evaluate using a calculator. (Refer to the Answers to Matched Problems
instruction book for your calculator to see how exponential forms
are evaluated.) 1. (A) 4 (B) - 4
(C) - 3 (D) Not a real number
89. 223>2 (E) 27
90. 155>4 5
2. (A) 2u (B) 29
16x2y 5 2 2 or 1 2
9
16x2y 5 2 2 2
91. 827-3>8 5
(C) 1> 2 13xy2 3 (D) 19u2 1>4
92. 103-3>4 (E) - 12x2 4>7 (F) 1x3 + y 3 2 1>3 (not x + y)
93. 37.097>3
3. (A) 27 (B) 81
94. 2.8768>5 12
(C) 10y 7>12 = 10 2 y 7 (D) 16y>x3
In Problems 95 and 96, evaluate each expression on a calcula- 18
(E) 2>x1>18 = 2> 2 x
tor and determine which pairs have the same value. Verify these
results algebraically. 4. (A) 10c13>4 - 2c (B) 14x + 19x1>2y 1>2 - 3y
-7>6
95. (A) 13 + 15 (B) 22 + 13 + 22 - 13 5. 2.5x - 2x -1
3
(C) 1 + 13 (D) 2 10 + 613 6. (A) x3 + y 3 (B) 2y (C) 2x
(E) 28 + 160 (F) 16 7. (A) 4b13ab (B) 31 16 - 132
3
96. (A) 222 + 15 (B) 18 (C) 1x + y2 1 1x + 1y2
(C) 13 + 17 (D) 23 + 18 + 23 - 18 8. (A) 1 (B)

1
(C)
1
16 2 + 1n 13 + h + 13
(E) 210 + 184 (F) 1 + 15
B.7 Quadratic Equations

■■ Solution by Square Root In this section we consider equations involving second-degree polynomials.
■■ Solution by Factoring
■■ Quadratic Formula DEFINITION Quadratic Equation
■■ Quadratic Formula and Factoring A quadratic equation in one variable is any equation that can be written in the form
■■ Other Polynomial Equations ax2 + bx + c = 0 a ∙ 0 Standard form
where x is a variable and a, b, and c are constants.
■■ Application: Supply and Demand
The equations
5x2 - 3x + 7 = 0 and 18 = 32t 2 - 12t
are both quadratic equations, since they are either in the standard form or can be
transformed into this form.
We restrict our review to finding real solutions to quadratic equations.
Solution by Square Root

The easiest type of quadratic equation to solve is the special form where the first-
degree term is missing:
ax2 ∙ c ∙ 0 a 3 0
The method of solution of this special form makes direct use of the square-root
property:
THEOREM 1 Square-Root Property

If a2 = b, then a = { 1b.
EXAMPLE 1 Square-Root Method Use the square-root property to solve each equation.
(A) x2 - 7 = 0 (B) 2x2 - 10 = 0
(C) 3x2 + 27 = 0 (D) 1x - 82 2 = 9
SOLUTION
(A) x2 - 7 = 0
x2 = 7 What real number squared is 7?
x = { 27 Short for 27 or - 27
(B) 2x2 - 10 = 0
2x2 = 10
x2 = 5 What real number squared is 5?
x = { 25
(C) 3x2 + 27 = 0
3x2 = - 27
x2 = - 9 What real number squared is - 9?
No real solution, since no real number squared is negative.
(D) 1x - 82 2 = 9
x - 8 = { 29
x - 8 = {3
x = 8 { 3 = 5 or 11
MATCHED PROBLEM 1 Use the square-root property to solve each equation.
2
(A) x - 6 = 0 (B) 3x2 - 12 = 0
(C) x2 + 4 = 0 (D) 1x + 52 2 = 1
Solution by Factoring
If the left side of a quadratic equation when written in standard form can be factored,
the equation can be solved very quickly. The method of solution by factoring rests on
a basic property of real numbers, first mentioned in Section B.1.
CONCEPTUAL INSIGHT
Theorem 2 in Section B.1 states that if a and b are real numbers, then ab = 0 if
and only if a = 0 or b = 0. To see that this property is useful for solving qua-
dratic equations, consider the following:
x2 - 4x + 3 = 0 (1)
1x - 121x - 32 = 0
x - 1 = 0 or x - 3 = 0
x = 1 or x = 3
You should check these solutions in equation (1).
If one side of the equation is not 0, then this method cannot be used. For
example, consider
x2 - 4x + 3 = 8 (2)
1x - 121x - 32 = 8
x - 1 ∙ 8 or x - 3 ∙ 8 ab = 8 does not imply
x = 9 or x = 11 that a = 8 or b = 8.
Verify that neither x = 9 nor x = 11 is a solution for equation (2).
EXAMPLE 2 Factoring Method Solve by factoring using integer coefficients, if possible.

(A) 3x2 - 6x - 24 = 0 (B) 3y 2 = 2y (C) x2 - 2x - 1 = 0
SOLUTION
(A) 3x2 - 6x - 24 = 0 Divide both sides by 3, since 3 is a factor
of each coefficient.
x2 - 2x - 8 = 0 Factor the left side, if possible.
1x - 421x + 22 = 0
x - 4 = 0 or x + 2 = 0
x = 4 or x = -2
(B) 3y 2 = 2y
3y 2 - 2y = 0 We lose the solution y = 0 if both sides are divided by y
y13y - 22 = 0 (3y 2 = 2y and 3y = 2 are not equivalent).
y = 0 or 3y - 2 = 0
3y = 2
2
y =
3
(C) x2 - 2x - 1 = 0
This equation cannot be factored using integer coefficients. We will solve this
type of equation by another method, considered below.
MATCHED PROBLEM 2 Solve by factoring using integer coefficients, if possible.
2
(A) 2x + 4x - 30 = 0 (B) 2x2 = 3x (C) 2x2 - 8x + 3 = 0
Note that an equation such as x2 = 25 can be solved by either the square-root

or the factoring method, and the results are the same (as they should be). Solve this
equation both ways and compare.
Also, note that the factoring method can be extended to higher-degree polyno-
mial equations. Consider the following:
x3 - x = 0
x1x2 - 12 = 0
x1x - 121x + 12 = 0
x = 0 or x - 1 = 0 or x + 1 = 0
Solution: x = 0, 1, - 1
Check these solutions in the original equation.
The factoring and square-root methods are fast and easy to use when they apply.
However, there are quadratic equations that look simple but cannot be solved by
either method. For example, as was noted in Example 2C, the polynomial in
x2 - 2x - 1 = 0
cannot be factored using integer coefficients. This brings us to the well-known and
widely used quadratic formula.
Quadratic Formula
There is a method called completing the square that will work for all quadratic equa-
tions. After briefly reviewing this method, we will use it to develop the quadratic
formula, which can be used to solve any quadratic equation.
The method of completing the square is based on the process of transforming a

quadratic equation in standard form,
ax2 + bx + c = 0
into the form
1x + A2 2 = B
where A and B are constants. Then, this last equation can be solved easily (if it has a
real solution) by the square-root method discussed above.
Consider the equation from Example 2C:
x2 - 2x - 1 = 0 (3)
Since the left side does not factor using integer coefficients, we add 1 to each side to
remove the constant term from the left side:
x2 - 2x = 1 (4)
Now we try to find a number that we can add to each side to make the left side a
square of a first-degree polynomial. Note the following square of a binomial:
1x + m2 2 = x2 + 2mx + m2
We see that the third term on the right is the square of one-half the coefficient of x
in the second term on the right. To complete the square in equation (4), we add the
square of one-half the coefficient of x, 1 - 22 2 2 = 1, to each side. (This rule works
only when the coefficient of x2 is 1, that is, a = 1.) Thus,
x2 - 2x ∙ 1 = 1 ∙ 1
The left side is the square of x - 1, and we write
1x - 12 2 = 2
What number squared is 2?
x - 1 = { 22
x = 1 { 22
And equation (3) is solved!
Let us try the method on the general quadratic equation
ax2 + bx + c = 0 a ∙ 0 (5)
and solve it once and for all for x in terms of the coefficients a, b, and c. We start by
multiplying both sides of equation (5) by 1/a to obtain
b c
x2 + x + = 0
a a
Add - c>a to both sides:
b c
x2 + x = -
a a
Now we complete the square on the left side by adding the square of one-half the
coefficient of x, that is, 1b>2a2 2 = b2 >4a2 to each side:
b b2 b2 c
x2 + x + = -
a 4a 2
4a 2 a
Writing the left side as a square and combining the right side into a single fraction,
we obtain
b 2 b2 - 4ac
ax + b =
2a 4a2
Now we solve by the square-root method:

b b2 - 4ac
x + = {
2a B 4a2
b 2b2 - 4ac Since { 24a2 = { 2a for
x = - { any real number a
2a 2a
When this is written as a single fraction, it becomes the quadratic formula:
Quadratic Formula
If ax2 + bx + c = 0, a ∙ 0, then
- b { 2b2 - 4ac
x =
2a
This formula is generally used to solve quadratic equations when the square-root
or factoring methods do not work. The quantity b2 - 4ac under the radical is called
the discriminant, and it gives us the useful information about solutions listed in
Table 1.
Table 1
b2 - 4ac ax2 + bx + c = 0
Positive Two real solutions
Zero One real solution
Negative No real solutions
EXAMPLE 3 Quadratic Formula Method Solve x2 - 2x - 1 = 0 using the quadratic

formula.
SOLUTION
x2 - 2x - 1 = 0
- b { 2b2 - 4ac
x = a = 1, b = - 2, c = -1
2a
- 1 - 22 { 21 - 22 2 - 41121 - 12
=
2112
2 { 18 2 { 2 12
= = = 1 { 12 ≈ - 0.414 or 2.414
2 2
CHECK
x2 - 2x - 1 = 0
When x = 1 + 22,
11 + 122 2 - 211 + 122 - 1 = 1 + 2 12 + 2 - 2 - 2 12 - 1 = 0
When x = 1 - 12,
11 - 12 2 2 - 211 - 122 - 1 = 1 - 2 12 + 2 - 2 + 2 12 - 1 = 0
MATCHED PROBLEM 3 Solve 2x2 - 4x - 3 = 0 using the quadratic formula.
If we try to solve x2 - 6x + 11 = 0 using the quadratic formula, we obtain
6 { 1-8
x =
2
which is not a real number. (Why?)
Quadratic Formula and Factoring

As in Section B.3, we restrict our interest in factoring to polynomials with integer
coefficients. If a polynomial cannot be factored as a product of lower-degree polyno-
mials with integer coefficients, we say that the polynomial is not factorable in the
integers.
How can you factor the quadratic polynomial x2 - 13x - 2,310? We start by
solving the corresponding quadratic equation using the quadratic formula:
x2 - 13x - 2,310 = 0
- 1 - 132 { 21 - 132 3 - 41121 - 2,3102

x =
2
- 1 - 132 { 29,409
x =
2
13 { 97
= = 55 or - 42
2
Now we write
x2 - 13x - 2,310 = 3x - 5543x - 1 - 4224 = 1x - 5521x + 422
Multiplying the two factors on the right produces the second-degree polynomial on
the left.
What is behind this procedure? The following two theorems justify and general-
ize the process:
THEOREM 2 Factorability Theorem

A second-degree polynomial, ax2 + bx + c, with integer coefficients can be
expressed as the product of two first-degree polynomials with integer coefficients
if and only if 2b2 - 4ac is an integer.
THEOREM 3 Factor Theorem

If r1 and r2 are solutions to the second-degree equation ax2 + bx + c = 0, then
ax2 + bx + c = a1x - r1 21x - r2 2
EXAMPLE 4 Factoring with the Aid of the Discriminant Factor, if possible, using integer
coefficients.
(A) 4x2 - 65x + 264 (B) 2x2 - 33x - 306
SOLUTION (A) 4x 2 - 65x + 264
Step 1 Test for factorability:
2b2 - 4ac = 21 - 652 2 - 414212642 = 1
Since the result is an integer, the polynomial has first-degree factors with
integer coefficients.
Step 2 Factor, using the factor theorem. Find the solutions to the corresponding
quadratic equation using the quadratic formula:
4x2 - 65x + 264 = 0
From step 1
- 1 - 652 { 1 33
2#4
x = = or 8
4
Thus,
33
4x2 - 65x + 264 = 4 ax - b 1x - 82
4
= 14x - 3321x - 82
(B) 2x2 - 33x - 306

Step 1 Test for factorability:
2b2 - 4ac = 21 - 332 2 - 41221 - 3062 = 13,537
Since 13,537 is not an integer, the polynomial is not factorable in the integers.
MATCHED PROBLEM 4 Factor, if possible, using integer coefficients.

2
(A) 3x - 28x - 464 (B) 9x2 + 320x - 144
Other Polynomial Equations

There are formulas that are analogous to the quadratic formula, but considerably
more complicated, that can be used to solve any cubic (degree 3) or quartic (degree 4)
polynomial equation. It can be shown that no such general formula exists for solving
quintic (degree 5) or polynomial equations of degree greater than five. Certain poly-
nomial equations, however, can be solved easily by taking roots.
EXAMPLE 5 Solving a Quartic Equation Find all real solutions to 6x4 - 486 = 0.
SOLUTION
6x4 - 486 = 0 Add 486 to both sides

6x4 = 486 Divide both sides by 6
x4 = 81 Take the 4th root of both sides
x = {3
MATCHED PROBLEM 5 Find all real solutions to 6x5 + 192 = 0.
Application: Supply and Demand

Supply-and-demand analysis is a very important part of business and economics.
In general, producers are willing to supply more of an item as the price of an item in-
creases and less of an item as the price decreases. Similarly, buyers are willing to buy
less of an item as the price increases, and more of an item as the price decreases. We
have a dynamic situation where the price, supply, and demand fluctuate until a price
is reached at which the supply is equal to the demand. In economic theory, this point
is called the equilibrium point. If the price increases from this point, the supply will
increase and the demand will decrease; if the price decreases from this point, the sup-
ply will decrease and the demand will increase.
EXAMPLE 6 Supply and Demand At a large summer beach resort, the weekly supply-and-
demand equations for folding beach chairs are
x 3
p = + Supply equation
140 4
5,670
p = Demand equation
x
The supply equation indicates that the supplier is willing to sell x units at a price of
p dollars per unit. The demand equation indicates that consumers are willing to buy
x units at a price of p dollars per unit. How many units are required for supply to
equal demand? At what price will supply equal demand?
SOLUTION Set the right side of the supply equation equal to the right side of the
demand equation and solve for x.
x 3 5,670
+ = Multiply by 140x, the LCD.
140 4 x
x2 + 105x = 793,800 Write in standard form.
x2 + 105x - 793,800 = 0 Use the quadratic formula.
- 105 { 21052 - 41121 - 793,8002

x =
2
x = 840 units
The negative root is discarded since a negative number of units cannot be produced
or sold. Substitute x = 840 back into either the supply equation or the demand
equation to find the equilibrium price (we use the demand equation).
5,670 5,670
p = = = $6.75
x 840
At a price of $6.75 the supplier is willing to supply 840 chairs and consumers are
willing to buy 840 chairs during a week.
MATCHED PROBLEM 6 Repeat Example 6 if near the end of summer, the supply-
and-demand equations are
x 1
p = - Supply equation
80 20
1,264
p = Demand equation
x
Exercises B.7
Find only real solutions in the problems below. If there are no real 17. 8x2 + 20x = 12 18. 9x2 - 6 = 15x
solutions, say so. 19. x2 = 1 - x 20. m2 = 1 - 3m
A Solve Problems 1–4 by the square-root method. 21. 2x2 = 6x - 3 22. 2x2 = 4x - 1
1. 2x2 - 22 = 0 2. 3m2 - 21 = 0 23. y 2 - 4y = - 8 24. x2 - 2x = - 3
3. 13x - 12 2 = 25 4. 12x + 12 2 = 16 25. 12x + 32 2 = 11 26. 15x - 22 2 = 7
Solve Problems 5–8 by factoring. 3 7

27. = p 28. x - = 0
p x
5. 2u2 - 8u - 24 = 0 6. 3x2 - 18x + 15 = 0
2 3 5 3
7. x2 = 2x 8. n2 = 3n 29. 2 - 2 = 30. 2 + = 2
m m u u
Solve Problems 9–12 by using the quadratic formula.
In Problems 31–38, factor, if possible, as the product of two first-
9. x2 - 6x - 3 = 0 10. m2 + 8m + 3 = 0
degree polynomials with integer coefficients. Use the quadratic
11. 3u2 + 12u + 6 = 0 12. 2x2 - 20x - 6 = 0 formula and the factor theorem.
B Solve Problems 13–30 by using any method. 31. x2 + 40x - 84 32. x2 - 28x - 128
2x2 3 33. x2 - 32x + 144 34. x2 + 52x + 208

13. = 5x 14. x2 = - x
3 4 35. 2x2 + 15x - 108 36. 3x2 - 32x - 140
15. 4u2 - 9 = 0 16. 9y 2 - 25 = 0 37. 4x2 + 241x - 434 38. 6x2 - 427x - 360
C 39. Solve A = P11 + r2 2 for r in terms of A and P; that is, 50. Supply and demand. An importer sells an automatic cam-
isolate r on the left side of the equation (with coefficient 1) era to outlets in a large city. During the summer, the weekly
and end up with an algebraic expression on the right side supply-and-demand equations are
involving A and P but not r. Write the answer using positive
x
square roots only. p = + 9 Supply equation
6
2
40. Solve x + 3mx - 3n = 0 for x in terms of m and n.
41. Consider the quadratic equation 24,840
p = Demand equation
2 x
x + 4x + c = 0
where c is a real number. Discuss the relationship between How many units are required for supply to equal demand? At
the values of c and the three types of roots listed in Table 1 on what price will supply equal demand?
page 605. 51. Interest rate. If P dollars are invested at 100r percent
compounded annually, at the end of 2 years it will grow to
42. Consider the quadratic equation
A = P11 + r2 2. At what interest rate will $484 grow to
x2 - 2x + c = 0 $625 in 2 years? (Note: If A = 625 and P = 484, find r.)
where c is a real number. Discuss the relationship between the 52. Interest rate. Using the formula in Problem 51, determine the
values of c and the three types of roots listed in Table 1 on interest rate that will make $1,000 grow to $1,210 in 2 years.
page 605.
53. Ecology. To measure the velocity v (in feet per second) of
In Problems 43–48, find all real solutions. a stream, we position a hollow L-shaped tube with one end
43. x3 + 8 = 0 under the water pointing upstream and the other end point-
ing straight up a couple of feet out of the water. The water
44. x3 - 8 = 0 will then be pushed up the tube a certain distance h (in feet)
45. 5x4 - 500 = 0 above the surface of the stream. Physicists have shown that
v2 = 64h. Approximately how fast is a stream flowing if
46. 2x3 + 250 = 0 h = 1 foot? If h = 0.5 foot?
47. x4 - 8x2 + 15 = 0 54. Safety research. It is of considerable importance to know
4 2
48. x - 12x + 32 = 0 the least number of feet d in which a car can be stopped,
including reaction time of the driver, at various speeds v (in
miles per hour). Safety research has produced the formula
d = 0.044v2 + 1.1v. If it took a car 550 feet to stop, estimate
Applications the car’s speed at the moment the stopping process was started.
49. Supply and demand. A company wholesales shampoo in a Answers to Matched Problems
particular city. Their marketing research department estab- 1. (A) { 16 (B) { 2
lished the following weekly supply-and-demand equations:
(C) No real solution (D) - 6, - 4
x 1 2. (A) - 5, 3 (B) 0, 32
p = + Supply equation (C) Cannot be factored using integer coefficients
450 2
3. 12 { 1102 >2
6,300 4. (A) Cannot be factored using integer coefficients
p = Demand equation
x (B) 19x - 42 1x + 362
How many units are required for supply to equal demand? 5. - 2
At what price per bottle will supply equal demand? 6. 320 chairs at $3.95 each
Appendix
C
C.1 Sequences, Series, and
Summation Notation
C.2 Arithmetic and Geometric
Special Topics
Sequences
C.3 Binomial Theorem
C.1 Sequences, Series, and Summation Notation

■■ Sequences If someone asked you to list all natural numbers that are perfect squares, you might
■■ Series and Summation Notation
begin by writing
1, 4, 9, 16, 25, 36
But you would soon realize that it is impossible to actually list all the perfect squares,
since there are an infinite number of them. However, you could represent this collec-
tion of numbers in several different ways. One common method is to write
1, 4, 9, c, n2, c n ∊ N
where N is the set of natural numbers. A list of numbers such as this is generally
called a sequence.
Sequences
Consider the function f given by
f1n2 = 2n + 1 (1)
where the domain of f is the set of natural numbers N. Note that
f112 = 3, f122 = 5, f132 = 7, c
The function f is an example of a sequence. In general, a sequence is a function with
domain a set of successive integers. Instead of the standard function notation used in
equation (1), sequences are usually defined in terms of a special notation.
The range value f1n2 is usually symbolized more compactly with a symbol such
as an. Thus, in place of equation (1), we write
an = 2n + 1
and the domain is understood to be the set of natural numbers unless something is
said to the contrary or the context indicates otherwise. The elements in the range are
A1
A2 APPENDIX C Special Topics
called terms of the sequence; a1 is the first term, a2 is the second term, and an is the
nth term, or general term.
a1 = 2112 + 1 = 3 First term
a2 = 2122 + 1 = 5 Second term
a3 = 2132 + 1 = 7 Third term
f
an = 2n + 1 General term
The ordered list of elements
3, 5, 7, c, 2n + 1, c
obtained by writing the terms of the sequence in their natural order with respect to the
domain values is often informally referred to as a sequence. A sequence also may be
represented in the abbreviated form 5an 6, where a symbol for the nth term is written
within braces. For example, we could refer to the sequence 3, 5, 7, c, 2n + 1, c
as the sequence 52n + 16.
If the domain of a sequence is a finite set of successive integers, then the
sequence is called a finite sequence. If the domain is an infinite set of successive
integers, then the sequence is called an infinite sequence. The sequence 52n + 16
discussed above is an infinite sequence.
EXAMPLE 1 Writing the Terms of a Sequence Write the first four terms of each sequence:
1 - 12 n
(A) an = 3n - 2 (B) e f
n
SOLUTION
1 -1 1
(A) 1, 4, 7, 10 (B) - 1, , ,
2 3 4
MATCHED PROBLEM 1 Write the first four terms of each sequence:

1 - 12 n
(A) an = - n + 3 (B) e f
2n
Now that we have seen how to use the general term to find the first few terms in
a sequence, we consider the reverse problem. That is, can a sequence be defined just
by listing the first three or four terms of the sequence? And can we then use these
initial terms to find a formula for the nth term? In general, without other information,
the answer to the first question is no. Many different sequences may start off with the
same terms. Simply listing the first three terms (or any other finite number of terms)
does not specify a particular sequence.
What about the second question? That is, given a few terms, can we find the
general formula for at least one sequence whose first few terms agree with the given
terms? The answer to this question is a qualified yes. If we can observe a simple pat-
tern in the given terms, we usually can construct a general term that will produce that
pattern. The next example illustrates this approach.
EXAMPLE 2 Finding the General Term of a Sequence Find the general term of a sequence
whose first four terms are
(A) 3, 4, 5, 6, c (B) 5, - 25, 125, - 625, c
SOLUTION
(A) Since these terms are consecutive integers, one solution is an = n, n Ú 3. If
we want the domain of the sequence to be all natural numbers, another solution
is bn = n + 2.
SECTION C.1 Sequences, Series, and Summation Notation A3
(B) Each of these terms can be written as the product of a power of 5 and a power
of - 1:
5 = 1 - 12 0 51 = a1
- 25 = 1 - 12 1 52 = a2
125 = 1 - 12 2 53 = a3
- 625 = 1 - 12 3 54 = a4
If we choose the domain to be all natural numbers, a solution is
an = 1 - 12 n - 1 5n
MATCHED PROBLEM 2 Find the general term of a sequence whose first four terms are
(A) 3, 6, 9, 12, c (B) 1, - 2, 4, - 8, c
In general, there is usually more than one way of representing the nth term of
a given sequence (see the solution of Example 2A). However, unless something is
stated to the contrary, we assume that the domain of the sequence is the set of natural
numbers N.
Series and Summation Notation

If a1, a2, a3, c, an, c is a sequence, the expression
a1 + a2 + a3 + g + an + g
is called a series. If the sequence is finite, the corresponding series is a finite series.
If the sequence is infinite, the corresponding series is an infinite series. We consider
only finite series in this section. For example,
1, 3, 5, 7, 9 Finite sequence
1 + 3 + 5 + 7 + 9 Finite series
Notice that we can easily evaluate this series by adding the five terms:
1 + 3 + 5 + 7 + 9 = 25
Series are often represented in a compact form called summation notation.
Consider the following examples:
ak = 3 + 4 + 5 + 6
6
2 2 2 2 2
k=3
= 9 + 16 + 25 + 36 = 86
a 14k + 12 = 14
2
# 0 + 12 + 14 # 1 + 12 + 14 # 2 + 12
k=0
= 1 + 5 + 9 = 15
In each case, the terms of the series on the right are obtained from the expression on
the number indicated below the summation sign g and ending with the number that
the left by successively replacing the summing index k with integers, starting with
appears above g . The summing index may be represented by letters other than k and
may start at any integer and end at any integer greater than or equal to the starting
integer. If we are given the finite sequence
1 1 1 1
, , , c, n
2 4 8 2
the corresponding series is
+ g + n = a j
n
1 1 1 1 1
+ +
2 4 8 2 j=1 2
where we have used j for the summing index.

EXAMPLE 3 Summation Notation Write
a k2 + 1
5
k
k=1
without summation notation. Do not evaluate the sum.

SOLUTION
a k 2 + 1 = 12 + 1 + 22 + 1 + 32 + 1 + 42 + 1 + 52 + 1
5
k 1 2 3 4 5
k=1
1 2 3 4 5
= + + + +
2 5 10 17 26
MATCHED PROBLEM 3 Write
a
5
k + 1
k=1 k
without summation notation. Do not evaluate the sum.
If the terms of a series are alternately positive and negative, we call the series an
alternating series. The next example deals with the representation of such a series.
EXAMPLE 4 Summation Notation Write the alternating series
1 1 1 1 1 1
- + - + -
2 4 6 8 10 12
using summation notation with
(A) The summing index k starting at 1
(B) The summing index j starting at 0
SOLUTION
(A) 1 - 12 k + 1 provides the alternation of sign, and 1> 12k2 provides the other part
of each term. So, we can write
= a
6 1 - 12 k + 1
1 1 1 1 1 1
- + - + -
2 4 6 8 10 12 k=1 2k
(B) 1 - 12 j provides the alternation of sign, and 1> 321j + 124 provides the other
part of each term. So, we can write
= a
1 1 1 1 1 1 5 1 - 12 j
- + - + -
2 4 6 8 10 12 j = 0 21j + 12
MATCHED PROBLEM 4 Write the alternating series

1 1 1 1
1 - + - +
3 9 27 81
using summation notation with
(A) The summing index k starting at 1
(B) The summing index j starting at 0
Summation notation provides a compact notation for the sum of any list of num-
bers, even if the numbers are not generated by a formula. For example, suppose that the
results of an examination taken by a class of 10 students are given in the following list:
87, 77, 95, 83, 86, 73, 95, 68, 75, 86
SECTION C.1 Sequences, Series, and Summation Notation A5
If we let a1, a2, a3, c, a10 represent these 10 scores, then the average test score is
given by
a
1 10 1
ak = 187 + 77 + 95 + 83 + 86 + 73 + 95 + 68 + 75 + 862
10 k = 1 10
1
= 18252 = 82.5
10
More generally, in statistics, the arithmetic mean a of a list of n numbers
a1, a2, c, an is defined as
n ka
1 n
a = ak
=1
EXAMPLE 5 Arithmetic Mean Find the arithmetic mean of 3, 5, 4, 7, 4, 2, 3, and 6.

SOLUTION
a
1 8 1 1
a = ak = 13 + 5 + 4 + 7 + 4 + 2 + 3 + 62 = 1342 = 4.25
8k = 1 8 8
MATCHED PROBLEM 5 Find the arithmetic mean of 9, 3, 8, 4, 3, and 6.
Exercises C.1
A Write the first four terms for each sequence in Problems 1–6. 19. 96, 65, 82, 74, 91, 88, 87, 91, 77, and 74
1. an = 2n + 3 2. an = 4n - 3 20. 100, 62, 95, 91, 82, 87, 70, 75, 87, and 82
n + 2 2n + 1
3. an = 4. an = B Write the first five terms of each sequence in Problems 21–26.
n + 1 2n
5. an = 1 - 32 n + 1 6. an = 1 - 14 2 n - 1 1 - 12 n + 1
21. an =
7. Write the 10th term of the sequence in Problem 1. 2n
22. an = 1 - 12 n 1n - 12 2
8. Write the 15th term of the sequence in Problem 2.
23. an = n31 + 1 - 12 n 4
9. Write the 99th term of the sequence in Problem 3.
1 - 1 - 12 n
10. Write the 200th term of the sequence in Problem 4. 24. an =
n
3 n-1
In Problems 11–16, write each series in expanded form without 25. an = a - b
2
summation notation, and evaluate.
1 n+1
26. an = a - b
11. a k 12. a k 2
6 5 2
k=1 k=1
In Problems 27–42, find the general term of a sequence whose
13. a 12k - 32 14. a 1 - 22 k

7 4 first four terms agree with the given terms.
k=4 k=0
27. - 2, - 1, 0, 1, c 28. 4, 5, 6, 7, c
29. 4, 8, 12, 16, c 30. - 3, - 6, - 9, - 12, c
15. a k 16. a k
3 4
1 1 31. 12, 34, 56, 78, c 32. 12, 23, 34, 45, c
k = 0 10 k=12
33. 1, - 2, 3, - 4, c 34. - 2, 4, - 8, 16, c
35. 1, - 3, 5, - 7, c 36. 3, - 6, 9, - 12, c
Find the arithmetic mean of each list of numbers in Problems
17–20. 37. 1, 2 4 8
5 , 25 , 125 , c 9 , 27 , 81 , c
38. 43, 16 64 256
39. x, x , x , x , c
2 3 4
40. 1, 2x, 3x2, 4x3, c
17. 5, 4, 2, 1, and 6
x2 x3 x4
18. 7, 9, 9, 2, and 4 41. x, - x3, x5, - x7, c 42. x, , , ,c
2 3 4
Write each series in Problems 43–50 in expanded form without 62. For each positive integer n, the sum of the series
summation notation. Do not evaluate.
1 1 1 1 - 12 n + 1
1- + - + g + is greater than or
43. a 1 - 12 k + 1 12k - 12 2 44. a
5 4 1 - 22 k + 1 2 3 4 n
1
k=1 k = 1 2k + 1 equal to .
2
45. a 46. a 2
5 7 1 - 12 k
2k
Some sequences are defined by a recursive formula—that is, a
k = 2 2k + 3 k=3k - k formula that defines each term of the sequence in terms of one or
more of the preceding terms. For example, if 5an 6 is defined by
47. a xk - 1 48. a xk + 1
5 3
1
k=1 k=1 k a1 = 1 and an = 2an - 1 + 1 for n Ú 2
then
49. a 50. a a2 = 2a1 + 1 = 2 # 1 + 1 = 3
4 1 - 12 kx2k + 1 4 1 - 12 kx2k
2k + 1 k = 0 2k + 2
a3 = 2a2 + 1 = 2 # 3 + 1 = 7
k=0
Write each series in Problems 51–54 using summation notation a4 = 2a3 + 1 = 2 # 7 + 1 = 15

with
(A) The summing index k starting at k = 1 and so on. In Problems 63–66, write the first five terms of each
sequence.
(B) The summing index j starting at j = 0
63. a1 = 2 and an = 3an - 1 + 2 for n Ú 2
51. 2 + 3 + 4 + 5 + 6
64. a1 = 3 and an = 2an - 1 - 2 for n Ú 2
52. 12 + 22 + 32 + 42
65. a1 = 1 and an = 2an - 1 for n Ú 2
1 1 1
53. 1 - + -
2 3 4 66. a1 = 1 and an = - 13 an - 1 for n Ú 2
1 1 1 1
54. 1 - 3 + 5 - 7 + 9
If A is a positive real number, the terms of the sequence defined by
Write each series in Problems 55–58 using summation notation
with the summing index k starting at k = 1. A 1 A
a1 = and an = aa + b for n Ú 2
2 2 n-1 an - 1
3 4 n + 1
55. 2 + + + g +
2 3 n can be used to approximate 1A to any decimal place accuracy
desired. In Problems 67 and 68, compute the first four terms of
1 1 1 this sequence for the indicated value of A, and compare the fourth
56. 1 + + 2 + g + 2
22 3 n term with the value of 1A obtained from a calculator.
67. A = 2 68. A = 6
1 1 1 1 - 12 n + 1
57. - + - g + 69. The sequence defined recursively by a1 = 1, a2 = 1,
2 4 8 2n
an = an - 1 + an - 2 for n Ú 3 is called the Fibonacci
58. 1 - 4 + 9 - g + 1 - 12 n + 1n2 sequence. Find the first ten terms of the Fibonacci sequence.
C In Problems 59–62, discuss the validity of each statement. If the 15 1 + 15 n

statement is true, explain why. If not, give a counterexample. 70. The sequence defined by bn = a b is related
5 2
59. For each positive integer n, the sum of the series to the Fibonacci sequence. Find the first ten terms (to three
decimal places) of the sequence 5bn 6 and describe the rela-
1 1 1
1 + + + g + is less than 4. tionship.
2 3 n
60. For each positive integer n, the sum of the series

1 1 1 1
+ + + g + n is less than 1. 1. (A) 2, 1, 0, - 1 (B) -21, 14, -81, 16
1
2 4 8 2
2. (A) an = 3n (B) an = 1 - 22 n - 1
6
61. For each positive integer n, the sum of the series 3. 2 + 32 + 43 + 54 + 5
4. (A) a (B) a
1 1 1 1 - 12 n + 1 5 1 - 12 k - 1 4 1 - 12 j
- + - g + is greater than or
2 4 8 2n k=1 3k - 1 j=0 3j
1
equal to . 5. 5.5
4
SECTION C.2 Arithmetic and Geometric Sequences A7
C.2 Arithmetic and Geometric Sequences

■■ Arithmetic and Geometric Sequences For most sequences, it is difficult to sum an arbitrary number of terms of the
■■ nth-Term Formulas
sequence without adding term by term. But particular types of sequences—arithmetic
sequences and geometric sequences—have certain properties that lead to convenient
■■ Sum Formulas for Finite Arithmetic and useful formulas for the sums of the corresponding arithmetic series and geometric
Series series.
■■ Sum Formulas for Finite Geometric
Series Arithmetic and Geometric Sequences
■■ Sum Formula for Infinite Geometric The sequence 5, 7, 9, 11, 13, c, 5 + 21n - 12, c, where each term after the
Series first is obtained by adding 2 to the preceding term, is an example of an arithmetic
■■ Applications sequence. The sequence 5, 10, 20, 40, 80, c, 5122 n - 1, c, where each term after
the first is obtained by multiplying the preceding term by 2, is an example of a geo-
metric sequence.
DEFINITION Arithmetic Sequence

A sequence of numbers
a1, a2, a3, c, an, c
is called an arithmetic sequence if there is a constant d, called the common
difference, such that
an - an - 1 = d
That is,
an = an - 1 + d for every n 7 1
DEFINITION Geometric Sequence

A sequence of numbers
a1, a2, a3, c, an, c
is called a geometric sequence if there exists a nonzero constant r, called a common
ratio, such that
an
= r
an - 1
That is,
an = ran - 1 for every n 7 1
EXAMPLE 1 Recognizing Arithmetic and Geometric Sequences Which of the following can
be the first four terms of an arithmetic sequence? Of a geometric sequence?
(A) 1, 2, 3, 5, c (B) - 1, 3, - 9, 27, c
(C) 3, 3, 3, 3, c (D) 10, 8.5, 7, 5.5, c
SOLUTION
(A) Since 2 - 1 ∙ 5 - 3, there is no common difference, so the sequence is not
an arithmetic sequence. Since 2>1 ∙ 3>2, there is no common ratio, so the
sequence is not geometric either.
(B) The sequence is geometric with common ratio - 3. It is not arithmetic.
(C) The sequence is arithmetic with common difference 0, and is also geometric
with common ratio 1.
(D) The sequence is arithmetic with common difference - 1.5. It is not geometric.
MATCHED PROBLEM 1 Which of the following can be the first four terms of an
arithmetic sequence? Of a geometric sequence?
(A) 8, 2, 0.5, 0.125, c (B) - 7, - 2, 3, 8, c (C) 1, 5, 25, 100, c
nth-Term Formulas
If 5an 6 is an arithmetic sequence with common difference d, then
a2 = a1 + d
a3 = a2 + d = a1 + 2d
a4 = a3 + d = a1 + 3d
This suggests that
THEOREM 1 nth Term of an Arithmetic Sequence

an = a1 + 1n - 12d for all n 7 1 (1)
Similarly, if 5an 6 is a geometric sequence with common ratio r, then

a2 = a1r
a3 = a2r = a1r 2
a4 = a3r = a1r 3
This suggests that
THEOREM 2 nth Term of a Geometric Sequence

an = a1r n - 1 for all n 7 1 (2)
EXAMPLE 2 Finding Terms in Arithmetic and Geometric Sequences

(A) If the 1st and 10th terms of an arithmetic sequence are 3 and 30, respectively,
find the 40th term of the sequence.
(B) If the 1st and 10th terms of a geometric sequence are 3 and 30, find the 40th
term to three decimal places.
SOLUTION
(A) First use formula (1) with a1 = 3 and a10 = 30 to find d:
an = a1 + 1n - 12d
a10 = a1 + 110 - 12d
30 = 3 + 9d
d = 3
Now find a40:
a40 = 3 + 39 # 3 = 120
(B) First use formula (2) with a1 = 3 and a10 = 30 to find r:
an = a1r n - 1
a10 = a1r 10 - 1
30 = 3r 9
r9 = 10
r = 101>9
Now find a40:
a40 = 31101>9 2 39 = 311039>9 2 = 64,633.041
MATCHED PROBLEM 2
(A) If the 1st and 15th terms of an arithmetic sequence are - 5 and 23, respec-
tively, find the 73rd term of the sequence.
(B) Find the 8th term of the geometric sequence
1 -1 1
, , ,c
64 32 16
Sum Formulas for Finite Arithmetic Series

If a1, a2, a3, c, an is a finite arithmetic sequence, then the corresponding series
a1 + a2 + a3 + g + an is called a finite arithmetic series. We will derive two
simple and very useful formulas for the sum of a finite arithmetic series. Let d be the
common difference of the arithmetic sequence a1, a2, a3, c, an and let Sn denote the
sum of the series a1 + a2 + a3 + g + an. Then
Sn = a1 + 1a1 + d2 + g + 3a1 + 1n - 22d4 + 3a1 + 1n - 12d4
Reversing the order of the sum, we obtain
Sn = 3a1 + 1n - 12d4 + 3a1 + 1n - 22d4 + g + 1a1 + d2 + a1
Something interesting happens if we combine these last two equations by addition
(adding corresponding terms on the right sides):
2Sn = 32a1 + 1n - 12d4 + 32a1 + 1n - 12d4 + g + 32a1 + 1n - 12d4 + 32a1 + 1n - 12d4
All the terms on the right side are the same, and there are n of them. Thus,
2Sn = n32a1 + 1n - 12d4
and we have the following general formula:
THEOREM 3 Sum of a Finite Arithmetic Series: First Form

n
Sn = 32a1 + 1n - 12d4 (3)
2
Replacing
n
3a1 + 1n - 12d4 3a + a1 + 1n - 12d4
in
2 1
by an from equation (1), we obtain a second useful formula for the sum:
THEOREM 4 Sum of a Finite Arithmetic Series: Second Form

n
Sn = 1a1 + an 2 (4)
2
EXAMPLE 3 Finding a Sum Find the sum of the first 30 terms in the arithmetic sequence:
3, 8, 13, 18, c
SOLUTION Use formula (3) with n = 30, a1 = 3, and d = 5:
S30 =
30
32 # 3 + 130 - 1254 = 2,265
2
MATCHED PROBLEM 3 Find the sum of the first 40 terms in the arithmetic sequence:
15, 13, 11, 9, c
EXAMPLE 4 Finding a Sum Find the sum of all the even numbers between 31 and 87.
SOLUTION First, find n using equation (1):
an = a1 + 1n - 12d
86 = 32 + 1n - 122
n = 28
Now find S28 using formula (4):
n
Sn = 1a + an 2
2 1
28
S28 = 132 + 862 = 1,652
2
MATCHED PROBLEM 4 Find the sum of all the odd numbers between 24 and 208.
Sum Formulas for Finite Geometric Series

If a1, a2, a3,c, an is a finite geometric sequence, then the corresponding series
a1 + a2 + a3 + g + an is called a finite geometric series. As with arithmetic
series, we can derive two simple and very useful formulas for the sum of a finite geo-
metric series. Let r be the common ratio of the geometric sequence a1, a2, a3, c, an
and let Sn denote the sum of the series a1 + a2 + a3 + g + an. Then
Sn = a1 + a1r + a1r 2 + g + a1r n - 2 + a1r n - 1
If we multiply both sides by r, we obtain
rSn = a1r + a1r 2 + a1r 3 + g + a1r n - 1 + a1r n
Now combine these last two equations by subtraction to obtain
rSn - Sn = 1a1r + a1r 2 + a1r 3 + g + a1r n - 1 + a1r n 2 - 1a1 + a1r + a1r 2 + g + a1r n - 2 + a1r n - 1 2
1r - 12Sn = a1r n - a1
Notice how many terms drop out on the right side. Solving for Sn, we have
THEOREM 5 Sum of a Finite Geometric Series: First Form

a1 1r n - 12
Sn = r ∙ 1 (5)
r - 1
Since an = a1r n - 1, or ran = a1r n, formula (5) also can be written in the form
THEOREM 6 Sum of a Finite Geometric Series: Second Form

ran - a1
Sn = r ∙ 1 (6)
r - 1
EXAMPLE 5 Finding a Sum Find the sum (to 2 decimal places) of the first ten terms of the
geometric sequence:
1, 1.05, 1.052, c
SOLUTION Use formula (5) with a1 = 1, r = 1.05, and n = 10:
a1 1r n - 12
Sn =
r - 1
111.0510 - 12
S10 =
1.05 - 1
0.6289
≈ ≈ 12.58
0.05
MATCHED PROBLEM 5 Find the sum of the first eight terms of the geometric
sequence:
100, 10011.082, 10011.082 2, c
Sum Formula for Infinite Geometric Series

Given a geometric series, what happens to the sum Sn of the first n terms as n in-
creases without stopping? To answer this question, let us write formula (5) in the
form
a1r n a1
Sn = -
r - 1 r - 1
It is possible to show that if - 1 6 r 6 1, then r n will approach 0 as n increases.
The first term above will approach 0 and Sn can be made as close as we please to the
second term, - a1 > 1r - 12 [which can be written as a1 > 11 - r2], by taking n suf-
ficiently large. So, if the common ratio r is between - 1 and 1, we conclude that the
sum of an infinite geometric series is
THEOREM 7 Sum of an Infinite Geometric Series

a1
S∞ = -1 6 r 6 1 (7)
1 - r
If r … - 1 or r Ú 1, then an infinite geometric series has no sum.
Applications
EXAMPLE 6 Loan Repayment A person borrows $3,600 and agrees to repay the loan in
monthly installments over 3 years. The agreement is to pay 1% of the unpaid bal-
ance each month for using the money and $100 each month to reduce the loan. What
is the total cost of the loan over the 3 years?
SOLUTION Let us look at the problem relative to a time line:
$3,600 $3,500 $3,400 ??? $200 $100 Unpaid balance
0 1 2 3 ??? 34 35 36 Months
0.01(3,600) 0.01(3,500) 0.01(3,400) ??? 0.01(300) 0.01(200) 0.01(100) 1% of

5 36 5 35 5 34 53 52 51 unpaid balance
The total cost of the loan is

1 + 2 + g + 34 + 35 + 36
The terms form a finite arithmetic series with n = 36, a1 = 1, and a36 = 36, so we
can use formula (4):
n
Sn = 1a + an 2
2 1
36
S36 = 11 + 362 = $666
2
We conclude that the total cost of the loan over 3 years is $666.
MATCHED PROBLEM 6 Repeat Example 6 with a loan of $6,000 over 5 years.
EXAMPLE 7 Economy Stimulation The government has decided on a tax rebate program to
stimulate the economy. Suppose that you receive $1,200 and you spend 80% of this,
and each of the people who receive what you spend also spend 80% of what they re-
ceive, and this process continues without end. According to the multiplier principle
in economics, the effect of your $1,200 tax rebate on the economy is multiplied
many times. What is the total amount spent if the process continues as indicated?
SOLUTION We need to find the sum of an infinite geometric series with the first
amount spent being a1 = 10.821$1,2002 = $960 and r = 0.8. Using formula (7),
we obtain
a1
S∞ =
1 - r
$960
= = $4,800
1 - 0.8
Assuming the process continues as indicated, we would expect the $1,200 tax re-
bate to result in about $4,800 of spending.
MATCHED PROBLEM 7 Repeat Example 7 with a tax rebate of $2,000.
Exercises C.2
A In Problems 1 and 2, determine whether the indicated sequence 7. 5 + 4.9 + 4.8 + g + 0.1
can be the first three terms of an arithmetic or geometric se-
1 1 1
quence, and, if so, find the common difference or common ratio 8. 1 - + - g-
and the next two terms of the sequence.
4 9 1002
1. (A) - 11, - 16, - 21, c (B) 2, - 4, 8, c B Let a1, a2, a3, c, an, cbe an arithmetic sequence. In Problems
(C) 1, 4, 9, c (D) 12, 16, 18
1
,c 9–14, find the indicated quantities.
2. (A) 5, 20, 100, c (B) - 5, - 5, - 5, c 9. a1 = 7; d = 4; a2 = ?; a3 = ?
(C) 7, 6.5, 6, c (D) 512, 256, 128, c 10. a1 = - 2; d = - 3; a2 = ?; a3 = ?
In Problems 3–8, determine whether the finite series is arithmetic, 11. a1 = 2; d = 4; a21 = ?; S31 = ?
geometric, both, or neither. If the series is arithmetic or geomet-
ric, find its sum. 12. a1 = 8; d = - 10; a15 = ?; S23 = ?
13. a1 = 18; a20 = 75; S20 = ?
3. a 1 - 12 k + 1 4. a 3
101 200
14. a1 = 203; a30 = 261; S30 = ?
k=1 k=1
Let a1, a2, a3, c, an, c be a geometric sequence. In Problems

1 1 1 15–28, find the indicated quantities.
5. 1 + + + g +
2 3 50
15. a1 = 3; r = - 2; a2 = ?; a3 = ?; a4 = ?
6. 3 - 9 + 27 - g - 320
16. a1 = 32; r = - 12; a2 = ?; a3 = ?; a4 = ?
SECTION C.3 Binomial Theorem A13
17. a1 = 1; a7 = 729; r = - 3; S7 = ? 43. Does there exist an infinite geometric series with a1 = 10
that has sum equal to 6? Explain.
18. a1 = 3; a7 = 2,187; r = 3; S7 = ?
44. Does there exist an infinite geometric series with a1 = 10
19. a1 = 100; r = 1.08; a10 = ?
that has sum equal to 5? Explain.
20. a1 = 240; r = 1.06; a12 = ?
21. a1 = 100; a9 = 200; r = ?
22. a1 = 100; a10 = 300; r = ?
Applications
23. a1 = 500; r = 0.6; S10 = ?; S∞ = ?
45. Loan repayment. If you borrow $4,800 and repay the loan
24. a1 = 8,000; r = 0.4; S10 = ?; S∞ = ?
by paying $200 per month to reduce the loan and 1% of the
25. S41 = a 13k + 32 = ? 26. S50 = a 12k - 32 = ?

41 50 unpaid balance each month for the use of the money, what is
the total cost of the loan over 24 months?
k=1 k=1
46. Loan repayment. If you borrow $5,400 and repay the loan
27. S8 = a 1 - 22 k - 1 = ? 28. S8 = a 2k = ?
8 8
by paying $300 per month to reduce the loan and 1.5% of the
k=1 k=1 unpaid balance each month for the use of the money, what is
29. Find the sum of all the odd integers between 12 and 68. the total cost of the loan over 18 months?
30. Find the sum of all the even integers between 23 and 97. 47. Economy stimulation. The government, through a subsidy
program, distributes $5,000,000. If we assume that each per-
31. Find the sum of each infinite geometric sequence (if it exists). son or agency spends 70% of what is received, and 70% of
(A) 2, 4, 8, c (B) 2, - 12, 18, c this is spent, and so on, how much total increase in spending
results from this government action? (Let a1 = $3,500,000.)
32. Repeat Problem 31 for:
(A) 16, 4, 1, c (B) 1, - 3, 9, c 48. Economy stimulation. Due to reduced taxes, a person has
an extra $1,200 in spendable income. If we assume that the
C 33. Find f112 + f122 + f132 + g + f1502 if person spends 65% of this on consumer goods, and the pro-
f1x2 = 2x - 3. ducers of these goods in turn spend 65% on consumer goods,
34. Find g112 + g122 + g132 + g + g11002 if and that this process continues indefinitely, what is the total
g1t2 = 18 - 3t. amount spent (to the nearest dollar) on consumer goods?
35. Find f112 + f122 + g + f1102 if f1x2 = 1 12 2 x. 49. Compound interest. If $1,000 is invested at 5% com-
pounded annually, the amount A present after n years forms
36. Find g112 + g122 + g + g1102 if g1x2 = 2x. a geometric sequence with common ratio 1 + 0.05 = 1.05.
37. Show that the sum of the first n odd positive integers is n2, Use a geometric sequence formula to find the amount A
using appropriate formulas from this section. in the account (to the nearest cent) after 10 years. After 20
years. (Hint: Use a time line.)
38. Show that the sum of the first n even positive integers is
n + n2, using formulas in this section. 50. Compound interest. If $P is invested at 100r% compounded
annually, the amount A present after n years forms a geomet-
39. If r = 1, neither the first form nor the second form for the ric sequence with common ratio 1 + r. Write a formula for
sum of a finite geometric series is valid. Find a formula for the amount present after n years. (Hint: Use a time line.)
the sum of a finite geometric series if r = 1.
40. If all of the terms of an infinite geometric series are less than Answers to Matched Problems
1, could the sum be greater than 1,000? Explain.
1. (A) The sequence is geometric with r = 14. It is not arithmetic.
41. Does there exist a finite arithmetic series with a1 = 1 and (B) The sequence is arithmetic with d = 5. It is not geometric.
an = 1.1 that has sum equal to 100? Explain. (C) The sequence is neither arithmetic nor geometric.
42. Does there exist a finite arithmetic series with a1 = 1 and 2. (A) 139 (B) - 2
an = 1.1 that has sum equal to 105? Explain. 3. - 960 4. 10,672 5. 1,063.66 6. $1,830 7. $8,000
C.3 Binomial Theorem

■■ Factorial The binomial form
■■ Development of the Binomial
1a + b2 n
Theorem where n is a natural number, appears more frequently than you might expect. The co-
efficients in the expansion play an important role in probability studies. The binomial
formula, which we will derive informally, enables us to expand 1a + b2 n directly for
n any natural number. Since the formula involves factorials, we digress for a moment
here to introduce this important concept.
Factorial
For n a natural number, n factorial, denoted by n!, is the product of the first n natural
numbers. Zero factorial is defined to be 1. That is,
DEFINITION n Factorial
n! = n # 1n - 12 # g # 2 # 1
1! = 1
0! = 1
It is also useful to note that n! can be defined recursively.
DEFINITION n Factorial—Recursive Definition
n! = n # 1n - 12! n Ú 1
EXAMPLE 1 Factorial Forms Evaluate.

(A) 5! = 5 # 4 # 3 # 2 # 1 = 120
8! 8 # 7!
(B) = = 8
7! 7!
10! 10 #9#8# 7!
(C) = = 720
7! 7!
MATCHED PROBLEM 1 Evaluate.

7! 8!
(A) 4! (B) (C)
6! 5!
The following formula involving factorials has applications in many areas of

mathematics and statistics. We will use this formula to provide a more concise form
for the expressions encountered later in this discussion.
THEOREM 1 For n and r integers satisfying 0 " r " n,

n!
nCr =
r!1n - r2!
EXAMPLE 2 Evaluating nCr

9! 9! 9 # 8 # 7!
2 # 7!
(A) 9C2 = = = = 36
2!19 - 22! 2!7!
5! 5! 5!
(B) 5C5 = = = = 1
5!15 - 52! 5!0! 5!
MATCHED PROBLEM 2 Find

(A) 5C2 (B) 6C0
SECTION C.3 Binomial Theorem A15
Development of the Binomial Theorem

Let us expand 1a + b2 n for several values of n to see if we can observe a pattern that
leads to a general formula for the expansion for any natural number n:
1a + b2 1 = a + b
1a + b2 2 = a2 + 2ab + b2
1a + b2 3 = a3 + 3a2b + 3ab2 + b3
1a + b2 4 = a4 + 4a3b + 6a2b2 + 4ab3 + b4
1a + b2 5 = a5 + 5a4b + 10a3b2 + 10a2b3 + 5ab4 + b5
CONCEPTUAL INSIGHT
1. The expansion of 1a + b2 n has 1n + 12 terms.
2. The power of a decreases by 1 for each term as we move from left to right.
3. The power of b increases by 1 for each term as we move from left to right.
4. In each term, the sum of the powers of a and b always equals n.
5. Starting with a given term, we can get the coefficient of the next term by mul-
tiplying the coefficient of the given term by the exponent of a and dividing by
the number that represents the position of the term in the series of terms. For
example, in the expansion of 1a + b2 4 above, the coefficient of the third term
is found from the second term by multiplying 4 and 3, and then dividing by 2
[that is, the coefficient of the third term = 14 # 32 >2 = 6].
We now postulate these same properties for the general case:
n n-1 n1n - 12 n - 2 2 n1n - 121n - 22 n - 3 3

1a + b2 n = an + a b + g + bn
1 # 2 1 # 2 # 3
a b + a b +
1
n! n! n! n! n!
= an + a n - 1b + a n - 2b 2 + a n - 3b 3 + g + bn
0!1n - 02! 1!1n - 12! 2!1n - 22! 3!1n - 32! n!1n - n2!
= nC0 an + nC1an - 1b + nC2 an - 2b2 + nC3 an - 3b3 + g + nCn bn
And we are led to the formula in the binomial theorem:
THEOREM 2 Binomial Theorem

For all natural numbers n,
1a + b2 n = nC0 an + nC1 an - 1b + nC2 an - 2b2 + nC3 an - 3b3 + g + nCn bn
EXAMPLE 3 Using the Binomial Theorem Use the binomial theorem to expand 1u + v2 6.
SOLUTION
1u + v2 6 = 6C0 u6 + 6C1 u5v + 6C2 u4v2 + 6C3 u3v3 + 6C4 u2v4 + 6C5 uv5 + 6C6 v6
= u6 + 6u5v + 15u4v2 + 20u3v3 + 15u2v4 + 6uv5 + v6
MATCHED PROBLEM 3 Use the binomial theorem to expand 1x + 22 5.

EXAMPLE 4 Using the Binomial Theorem Use the binomial theorem to find the sixth term in
the expansion of 1x - 12 18.
13 18!
SOLUTION Sixth term = 18C5 x 1 - 12 5 = x13 1 - 12
5!118 - 52!
= - 8,568x13
MATCHED PROBLEM 4 Use the binomial theorem to find the fourth term in the
20
expansion of 1x - 22 .
Exercises C.3
A In Problems 1–20, evaluate each expression. C 33. Show that nC0 = nCn for n Ú 0.
10!
1. 6! 2. 7! 3. 34. Show that nCr = nCn - r for n Ú r Ú 0.
9!
20! 12! 10! 35. The triangle shown here is called Pascal’s triangle. Can you
4. 5. 6. guess what the next two rows at the bottom are? Compare
19! 9! 6!
these numbers with the coefficients of binomial expansions.
5! 7! 6!
7. 8. 9.
2!3! 3!4! 5!16 - 52! 1
7! 20! 52! 1 1
10. 11. 12.
4!17 - 42! 3!17! 50!2! 1 2 1
1 3 3 1
13. 5C3 14. 7C3 15. 6C5 16. 7C4
1 4 6 4 1
17. 5C0 18. 5C5 19. 18C15 20. 18C3
36. Explain why the sum of the entries in each row of Pascal’s
B Expand each expression in Problems 21–26 using the binomial triangle is a power of 2. (Hint: Let a = b = 1 in the bino-
theorem.
mial theorem.)
21. 1a + b2 4 22. 1m + n2 5
37. Explain why the alternating sum of the entries in each row of
23. 1x - 12 6 24. 1u - 22 5 Pascal’s triangle (e.g., 1 - 4 + 6 - 4 + 1) is equal to 0.
n - r + 1
25. 12a - b2 5 26. 1x - 2y2 5 38. Show that nCr = r n Cr - 1 for n Ú r Ú 1.
Find the indicated term in each expansion in Problems 27–32. 39. Show that nCr - 1 + n Cr = n + 1Cr for n Ú r Ú 1.
27. 1x - 12 18; 5th term 28. 1x - 32 20; 3rd term Answers to Matched Problems
15
29. 1p + q2 ; 7th term 15
30. 1p + q2 ; 13th term 1. (A) 24 (B) 7 (C) 336
2. (A) 10 (B) 1
31. 12x + y2 12; 11th term 32. 12x + y2 12; 3rd term 3. x5 + 10x4 + 40x3 + 80x2 + 80x + 32
4. - 9,120x17
Appendix
D
Area under the
Standard Normal Curve
f (x)
Area
corresponding
to z
m m 1 zs x
0 z z
Area under the Standard Normal Curve

(Table Entries Represent the Area under the Standard Normal Curve from 0 to z, z # 0)
z .00 .01 .02 .03 .04 .05 .06 .07 .08 .09
0.0 0.0000 0.0040 0.0080 0.0120 0.0160 0.0199 0.0239 0.0279 0.0319 0.0359
0.1 0.0398 0.0438 0.0478 0.0517 0.0557 0.0596 0.0636 0.0675 0.0714 0.0753
0.2 0.0793 0.0832 0.0871 0.0910 0.0948 0.0987 0.1026 0.1064 0.1103 0.1141
0.3 0.1179 0.1217 0.1255 0.1293 0.1331 0.1368 0.1406 0.1443 0.1480 0.1517
0.4 0.1554 0.1591 0.1628 0.1664 0.1700 0.1736 0.1772 0.1808 0.1844 0.1879
0.5 0.1915 0.1950 0.1985 0.2019 0.2054 0.2088 0.2123 0.2157 0.2190 0.2224
0.6 0.2257 0.2291 0.2324 0.2357 0.2389 0.2422 0.2454 0.2486 0.2517 0.2549
0.7 0.2580 0.2611 0.2642 0.2673 0.2704 0.2734 0.2764 0.2794 0.2823 0.2852
0.8 0.2881 0.2910 0.2939 0.2967 0.2995 0.3023 0.3051 0.3078 0.3106 0.3133
0.9 0.3159 0.3186 0.3212 0.3238 0.3264 0.3289 0.3315 0.3340 0.3365 0.3389
1.0 0.3413 0.3438 0.3461 0.3485 0.3508 0.3531 0.3554 0.3577 0.3599 0.3621
1.1 0.3643 0.3665 0.3686 0.3708 0.3729 0.3749 0.3770 0.3790 0.3810 0.3830
1.2 0.3849 0.3869 0.3888 0.3907 0.3925 0.3944 0.3962 0.3980 0.3997 0.4015
1.3 0.4032 0.4049 0.4066 0.4082 0.4099 0.4115 0.4131 0.4147 0.4162 0.4177
1.4 0.4192 0.4207 0.4222 0.4236 0.4251 0.4265 0.4279 0.4292 0.4306 0.4319
1.5 0.4332 0.4345 0.4357 0.4370 0.4382 0.4394 0.4406 0.4418 0.4429 0.4441
1.6 0.4452 0.4463 0.4474 0.4484 0.4495 0.4505 0.4515 0.4525 0.4535 0.4545
1.7 0.4554 0.4564 0.4573 0.4582 0.4591 0.4599 0.4608 0.4616 0.4625 0.4633
1.8 0.4641 0.4649 0.4656 0.4664 0.4671 0.4678 0.4686 0.4693 0.4699 0.4706
1.9 0.4713 0.4719 0.4726 0.4732 0.4738 0.4744 0.4750 0.4756 0.4761 0.4767
2.0 0.4772 0.4778 0.4783 0.4788 0.4793 0.4798 0.4803 0.4808 0.4812 0.4817
2.1 0.4821 0.4826 0.4830 0.4834 0.4838 0.4842 0.4846 0.4850 0.4854 0.4857
2.2 0.4861 0.4864 0.4868 0.4871 0.4875 0.4878 0.4881 0.4884 0.4887 0.4890
2.3 0.4893 0.4896 0.4898 0.4901 0.4904 0.4906 0.4909 0.4911 0.4913 0.4916
2.4 0.4918 0.4920 0.4922 0.4925 0.4927 0.4929 0.4931 0.4932 0.4934 0.4936
2.5 0.4938 0.4940 0.4941 0.4943 0.4945 0.4946 0.4948 0.4949 0.4951 0.4952
2.6 0.4953 0.4955 0.4956 0.4957 0.4959 0.4960 0.4961 0.4962 0.4963 0.4964
2.7 0.4965 0.4966 0.4967 0.4968 0.4969 0.4970 0.4971 0.4972 0.4973 0.4974
2.8 0.4974 0.4975 0.4976 0.4977 0.4977 0.4978 0.4979 0.4979 0.4980 0.4981
2.9 0.4981 0.4982 0.4982 0.4983 0.4984 0.4984 0.4985 0.4985 0.4986 0.4986
3.0 0.4987 0.4987 0.4987 0.4988 0.4988 0.4989 0.4989 0.4989 0.4990 0.4990
3.1 0.4990 0.4991 0.4991 0.4991 0.4992 0.4992 0.4992 0.4992 0.4993 0.4993
3.2 0.4993 0.4993 0.4994 0.4994 0.4994 0.4994 0.4994 0.4995 0.4995 0.4995
3.3 0.4995 0.4995 0.4995 0.4996 0.4996 0.4996 0.4996 0.4996 0.4996 0.4997
3.4 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4997 0.4998
3.5 0.4998 0.4998 0.4998 0.4998 0.4998 0.4998 0.4998 0.4998 0.4998 0.4998
3.6 0.4998 0.4998 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999
3.7 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999
3.8 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999 0.4999
3.9 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000
610
ANSWERS
Diagnostic Prerequisite Test 43. x = - 5 45. x = - 6 47. All real numbers 49. All real
numbers except - 4 51. x … 7 53. Yes; all real numbers 55. No; for
Section references are provided in parentheses following each answer example, when x = 0, y = {2 57. Yes; all real numbers except 0
to guide students to the specific content in the book where they can 59. No; when x = 1, y = {1 61. 25x2 - 4 63. x2 + 6x + 5
find help or remediation. 65. x4 - 4 67. x - 4 69. h2 - 4 71. 4h + h2 73. 4h + h2
1. (A) (y + z)x (B) (2 + x) + y (C) 2x + 3x (B.1) 75. (A) 4x + 4h - 3 (B) 4h (C) 4
2. x3 - 3x2 + 4x + 8 (B.2) 3. x3 + 3x2 - 2x + 12 (B.2) 77. (A) 4x2 + 8xh + 4h2 - 7x - 7h + 6 (B) 8xh + 4h2 - 7h (C) 8x + 4h - 7
4. - 3x5 + 2x3 - 24x2 + 16 (B.2) 5. (A) 1 (B) 1 (C) 2 79. (A) 20x + 20h - x2 - 2xh - h2 (B) 20h - 2xh - h2 (C) 20 - 2x - h
(D) 3 (B.2) 6. (A) 3 (B) 1 (C) - 3 (D) 1 (B.2) 7. 14x2 - 30x (B.2) 50
81. P1w2 = 2w + , w 7 0 83. A1l2 = l150 - l2, 0 6 l 6 50
8. 6x2 - 5xy - 4y 2 (B.2) 9. 1x + 221x + 52 (B.3) w
10. x1x + 321x - 52 (B.3) 11. 7>20 (B.1) 12. 0.875 (B.1) 85. (A) p(x) (B) $54; $42
13. (A) 4.065 * 1012 (B) 7.3 * 10 -3 (B.5) 14. (A) 255,000,000 $100
(B) 0,000 406 (B.5) 15. (A) T (B) F (B.1) 16. 0 and - 3 are two
examples of infinitely many. (B.1) 17. 6x5y 15 (B.5) 18. 3u4 >v2 (B.5)
19. 6 * 102 (B.5) 20. x6/y 4 (B.5) 21. u7>3 (B.6) 22. 3a2 >b (B.6)
20 x
5 a 2 + b2 a 2 - c2
23. 9 (B.5) 24. x + 2x1>2y 1>2 + y (B.6) 25. (B.4) 26. 87. (A) R1x2 = 175 - 3x2x, 1 … x … 20 (B) x R(x)
ab abc
5
y 1 -1 xy 1 72
(B.4) 27. (B.4) 28. 2 (B.4) 29. (B.4) 30. (B.6)
x xy 7(7 + h) y-x 4 252
31. (A) Subtraction (B) Commutative ( + ) (C) Distributive (D) Associative ( # ) 8 408
(E) Negatives (F) Identity ( + ) (B.1) 32. (A) 6 (B) 0 (B.1) 12 468
33. 4x = x - 4; x = - 4>3 (A.1) 34. - 15>7 (A.2) 35. (4/7, 0) (A.2) 16 432
36. (0, - 4) (A.2) 37. x = 0, 5 (B.7) 38. x = { 17 (B.7)
20 300
39. x = - 4, 5 (B.7) 40. x = 1, 16 (B.7)
R(x)
(C) $500 89. (A) P1x2 = 59x - 3x2 - 125, 1 … x … 20
Chapter 1
Exercises 1.1 20 x
1. y 3. y
5 5 (B) x P(x) 75 - w
(C) P(x) 91. v = ;
1 -69 $200 15 + w
5 x 10 x 4 63 1.9032 cm/sec
8 155 15 x
5. y 7. y 12 151 2200
50
4 16 51
20 -145
5 x 3 x
92. (A) v = 0.625s + 0.125; 44.4% (B) s = 1.6v - 0.2; 61.6%
9. A function 11. Not a function 13. A function Exercises 1.2

15. A function 17. Not a function 19. A function 1. Domain: all real numbers; range: [ - 4, ∞] 3. Domain:all real numbers; range:
21. Linear 23. Linear 25. Neither 27. Constant all real numbers 5. Domain: [0, ∞ 2; range: 1 - ∞, 8] 7. Domain: all real
29. y 31. y numbers; range: all real numbers 9. Domain: all real numbers; range: [9, ∞ 2
5 10
11. 13.
5 5
5 x 5 x
5 x 5 x
33. y 35. y
50 10
15. 17.
5 5
5 x 10 x
2 x 5 x
37. y 39. y = 0 41. y = - 2

5
5 x
A-1
A-2 Answers
19. 21. 55. The graph of the basic function y = x is reflected in the x axis and verti-
5 5 cally shrunk by a factor of 0.5. Equation: y = - 0.5 x . 57. The graph of
the basic function y = x2 is reflected in the x axis and vertically stretched by a
5 x 5 x 3
factor of 2. Equation: y = - 2x2. 59. The graph of the basic function y = 2 x
is reflected in the x axis and vertically stretched by a factor of 3. Equation:
3
y = -32 x. 61. Reversing the order does not change the result. 63. Re-
23. 25. versing the order can change the result. 65. Reversing the order can change
5 5
the result. 67. (A) The graph of the basic function y = 1x is reflected in
the x axis, vertically expanded by a factor of 4, and shifted up 115 units. (B)
5 x 5 x
p(x) 69. (A) The graph of the basic function y = x3 is verti-
$100 cally contracted by a factor of 0.000 48 and
27. The graph of g1x2 = - x + 3 is the graph of y = x reflected in shifted right 500 units and up 60,000 units.
$50
the x axis and shifted 3 units to the left. g(x)
10 0 100 200 x
10 x
1.34 + 0.32x if 0 … x … 11
(B) C(x) 71. (A) V1x2 = e
1.01 + 0.35x if x 7 11
29. The graph of f1x2 = 1x - 42 2 - 3 is the graph of y = x2 shifted 4 units $100,000 V(x)
to the right and 3 units down. f(x)
(B) 10
$50,000
10
0 500 1000 x
10 x
20 x
31. The graph of f1x2 = 7 - 1x is the graph of y = 1x reflected in the 0.002x if 0 … x … 15

x axis and shifted 7 units up. f(x) 73. (A) T1x2 = • 0.006x - 0.06 if 15 … x … 60
10 0.010x - 0.30 if x 7 60
T(x)
(B) (C) 0.12 tax units; 0.60 tax units
10 x 0.8
33. The graph of h1x2 = - 3 x is the graph of y = x reflected in the

x axis and vertically stretched by a factor of 3. h(x) 100 x
10
75. (A) The graph of the basic (B) w(x)

10 x function y = x is vertically stretched 250
by a factor of 5.5 and shifted down 200
150
220 units. 100
35. The graph of the basic function y = x2 is shifted 2 units to the left and 50
3 units down. Equation: y = 1x + 22 2 - 3. 37. The graph of the basic 0 50 100 x
function y = x2 is reflected in the x axis and shifted 3 units to the right and 77. (A) The graph of the basic func- (B) v(x)
2 units up. Equation: y = 2 - 1x - 32 2. 39. The graph of the basic tion y = 1x is vertically stretched by 100
function y = 1x is reflected in the x axis and shifted 4 units up. Equation: a factor of 7.08.
y = 4 - 1x. 41. The graph of the basic function y = x3 is shifted 2 units 50
to the left and 1 unit down. Equation: y = 1x + 22 3 - 1.
0 50 100 x
43. g1x2 = 1x + 3 + 2 45. g1x2 = x + 3
g(x) g(x) Exercises 1.3
5 5
1. f1x2 = 1x - 52 2 - 25 3. f1x2 = 1x + 102 2 - 50
5 1 2 1
x 5 x 5. f1x2 = - 21x - 12 2 - 3 7. f1x2 = 2 a x + b +
2 2
9. The graph of f1x2 is the graph of y = x2 shifted right 2 units and down 1
unit. 11. The graph of m1x2 is the graph of y = x2 reflected in the x axis,
47. g1x2 = - 1x - 22 3 - 1 49. f(x)
5 then shifted right 5 units and up 9 units. 13. (A) m (B) g (C) f
g(x)
5 (D) n 15. (A) x int.: 1, 3; y int.: - 3 (B) Vertex: (2, 1) (C) Max.: 1
5 x (D) Range: y … 1 or 1 - ∞, 14 17. (A) x int.: - 3, - 1; y int.: 3
5 x (B) Vertex: 1 - 2, - 12 (C) Min.: - 1 (D) Range: y Ú - 1 or 3 - 1, ∞ 2
19. (A) x int.: 3 { 12; y int.: - 7 (B) Vertex: 13, 22 (C) Max.: 2
(D) Range: y … 2 or 1 - ∞, 24 21. (A) x int.: - 1 { 12; y int.: - 1
51. h(x) 53. h(x) (B) Vertex: 1 - 1, - 22 (C) Min.: - 2 (D) Range: y Ú - 2 or 3 - 2, ∞ 2
30 100
20 x 80 x
Answers A-3
23. y = - 3x - 1 - 224 2 + 5 or y = - 1x + 22 2 + 5 75. x = 0.14 cm 77. 10.6 mph

25. y = 1x - 12 2 - 3 27. Vertex form: 1x - 42 2 - 4 (A) x int.: 2, 6;
y int.: 12 (B) Vertex: 14, - 42 (C) Min.: - 4 (D) Range: y Ú - 4
or 3 - 4, ∞ 2 29. Vertex form: - 41x - 22 2 + 1 (A) x int.: 1.5, 2.5; y int.:
- 15 (B) Vertex: 12, 12 (C) Max.: 1 (D) Range: y … 1 or 1 - ∞, 14
31. Vertex form: 0.51x - 22 2 + 3 (A) x int.: none; y int.: 5
(B) Vertex: 12, 32 (C) Min.: 3 (D) Range: y Ú 3 or 33, ∞ 2
Exercises 1.4
33. (A) - 4.87, 8.21 (B) - 3.44, 6.78 (C) No solution 35. 651.0417
37. g1x2 = 0.251x - 32 2 - 9.25 (A) x int.: - 3.08, 9.08; y int.: - 7 1. (A) 1 (B) - 3 (C) 21 3. (A) 2 (B) - 5, - 4 (C) 20 5. (A) 10
(B) Vertex: 13, - 9.252 (C) Min.: - 9.25 (D) Range: y Ú - 9.25 (B) None (C) 9 7. (A) 5 (B) 0, - 6 (C) 0 9. (A) 11 (B) - 5, - 2, 5
or 3 - 9.25, ∞ 2 39. f1x2 = - 0.121x - 42 2 + 3.12 (A) x int.: - 1.1, 9.1; (C) - 12,800 11. (A) 4 (B) Negative 13. (A) 5 (B) Negative
or 1 - ∞, 3.124 41. 1 - ∞, - 52 h 13, ∞ 2 43. 3 - 3, 24

y int.: 1.2 (B) Vertex: 14, 3.122 (C) Max.: 3.12 (D) Range: y … 3.12 15. (A) 1 (B) Negative 17. (A) 6 (B) Positive 19. 10 21. 1
23. (A) x int.: - 2; y int.: - 1 (B) Domain: all real numbers except 2
45. x = - 5.37, 0.37 47. - 1.37 6 x 6 2.16 49. x … - 0.74 or x Ú 4.19 (C) Vertical asymptote: x = 2; horizontal asymptote: y = 1
51. Axis: x = - 4; vertex: ( - 4, - 9); range: y … - 9 or ( - ∞, - 9); no x (D) f(x) 25. (A) x int.: 0; y int.: 0 (B) Domain: all real num-
intercept. 53. (A) y (B) 1.64, 7.61 10 bers except - 2 (C) Vertical asymptote: x = - 2;
15 g
horizontal asymptote: y = 3 (D) f(x)
10 x 10
f
10 x
10 x
(C) 1.64 6 x 6 7.61 (D) 0 … x 6 1.64 or 7.61 6 x … 10 27. (A) x int.: 2; y int.: - 1 (B) Domain: all real numbers except 4 (C) Vertical
55. (A) y (B) 1.10, 5.57 (C) 1.10 6 x 6 5.57 asymptote: x = 4; horizontal asymptote: y = - 2 (D) f(x)
15 g 10
(D) 0 … x 6 1.10 or 5.57 6 x … 8
10
x
f
10 x
65. (A) 29. (A)
x 28 30 32 34 36
Mileage 45 52 55 51 47
f(x) 45.3 51.8 54.2 52.4 46.5
(B) y (C) f1312 = 53.50 thousand miles;
50 (D) f1352 = 49.95 thousand miles;
40
30 (B)
20
10
0 28 32 36 x
69. (A) R(x) (B) 12.5 (12,500,000 chips);

500
$468,750,000 (C) $37.50
20 x 31. (A)
71. (A) y (B) 2,415,000 chips and 17,251,000
500
C chips (C) Loss: 1 … x 6 2.415
R
or 17.251 6 x … 20; profit:
2.415 6 x 6 17.251
20 x
73. (A) P1x2 = 59x - 3x2 - 125 y

(B)
500 C
R
15 x
2200 P
(C) Intercepts and break-even points: 2,415,000 chips and 17,251,000 chips
(D) Maximum profit is $165,083,000 at a production level of 9,833,000 chips.
This is much smaller than the maximum revenue of $468,750,000.
4 1
33. y = 35. y = 37. y = 0 39. None 41. x = - 1, x = 1,
5 4
x = - 3, x = 3 43. x = 5 45. x = - 7, x = 7 47. (A) x int.: 0; y int.:
0 (B) Vertical asymptotes: x = - 2, x = 3; horizontal asymptote: y = 2
A-4 Answers
(C) f(x) (D) 63. (A) (B) 1.7 lb 65. (A) 0.06 cm/sec
10
10 x
2 (B) v(x) 67. (A) (B) 5.5

49. (A) x int.: { 13; y int.: - (B) Vertical asymptotes: x = - 3, x = 3; 5
3
horizontal asymptote: y = - 2
10
(C) f(x) (D)
10
100 x
10 210 10
x
Exercises 1.5
210 1. (A) k (B) g (C) h (D) f 3. y
30
51. (A) x int.: 6; y int.: - 4 (B) Vertical asymptotes: x = - 3, x = 2; hori-
zontal asymptote: y = 0
10
(C) f(x) (D)
10
24 4 x
210 10
10 y
x 5. 7. 9. y
30 f(x)
24 4 x 24 4 x
210
53. f1x2 = x2 + 2x - 3 55. f1x2 = x3 + x2 - 2x

180x + 200 24 4 x 230 220
57. (A) C1x2 = 180x + 200 (B) C1x2 =
x
(C) C(x) (D) $180 per board 11. The graph of g is the graph of f reflected in the x axis. 13. The graph of
500
g is the graph of f shifted 1 unit to the left. 15. The graph of g is the graph of
f shifted 1 unit up. 17. The graph of g is the graph of f vertically stretched by
a factor of 2 and shifted to the left 2 units.
x
19. (A) y (B) y (C) y
0 30 5 5 5
2,500 + 175n + 25n2 C(n)
59. (A) C1n2 = (B) 1,500
n 5 x 5 x 5 x
(D) 21. f(t) 23.

y 10 y
0 20 n 5 20
(C) 10 yr; $675.00 per year (D) 10 yr; $675.00 per year
4 x 25 5
x
230 30 t 210
25. y 27. a = 1, - 1 29. x = 48 31. x = - 3, 7

20
33. x = - 9 35. x = 3, 19 37. x = - 4, - 3
39. x = - 7 41. x = - 2, 2 43. x = 1>4
45. No solution
0.000481x - 5002 3 + 60,000 24 4 x
61. (A) C1x2 =
x h(x) N
47. 49. 100 51. $129,239.88
25 x
(B) (C) 750 cases per month; 53. (A) $2,633.56
$90 per case (B) $7,079.54
20.5 55. $10,706
5 t
57. (A) $10,491.24 (B) $10,509.45 (C) $10,526.76 59. N approaches

N
2 as t increases without bound. 2
50 t
Answers A-5
61. (A) (B) 9.94 billion 63. (A) 10% Chapter 1 Review Exercises
(B) 1% 65. (A) P = 12e0.0402x
1. y (1.1) 2. y (1.1) 3. y (1.1)
(B) 17.9 million 10 5 5
67. (A) P = 127e - 0.0016x
(B) 124 million 5 x 5 x 5 x
4. (A) Not a function (B) A function (C) A function (D) Not a

Exercises 1.6
3 function (1.1) 5. (A) - 2 (B) - 8 (C) 0 (D) Not defined (1.1)
1. 27 = 33 3. 100 = 1 5. 8 = 43>2 7. log7 49 = 2 9. log4 8 =
2 6. v = ln u (1.6) 7. ek = m (1.6) 8. x = ln k (1.6) 9. z = log m (1.6)
11. logb A = u 13. 6 15. - 5 17. 7 19. - 3 21. Not defined
p 10. x = 64 (1.6) 11. z = 7 (1.6) 12. x = 3 (1.6) 13. x = 2.116 (1.6)
23. logb P - logb Q 25. 5 logb L 27. q 29. x = 1>10 31. b = 4
14. y = 5.298 (1.6) 15. z = 141.254 (1.6) 16. x = 0.318 (1.6)
33. y = - 3 35. b = 1>3 37. x = 8 39. False 41. True 43. True
17. (A) y = 4 (B) x = 0 (C) y = 1 (D) x = - 1 or 1 (E) y = - 2
45. False 47. x = 2 49. x = 8 51. x = 7 53. No solution
(F) x = - 5 or 5 (1.1)
55. y 57. The graph of y = log2 1x - 22 is the graph of
5
18. (A) y (B) y (C) y
y = log2 x shifted to the right 2 units. 5 5 5
59. Domain: 1 - 1, ∞2; range: all real numbers
10 x 61. (A) 3.547 43 (B) - 2.160 32 (C) 5.626 29 5 x 5 x 5 x
(D) - 3.197 04 63. (A) 13.4431 (B) 0.0089
(C) 16.0595 (D) 0.1514 65. 1.0792 67. 1.6923
69. 18.3559 71. Increasing: 10, ∞ 2 y (D) y (1.2) 19. f1x2 = - 1x - 22 2 + 4. The
3 5 graph of f1x2 is the graph of
y = x2 reflected in the x axis, then
5 x 5 x shifted right 2 units and up 4 units.
(1.2) 20. (A) g (B) m (C) n
(D) f (1.2, 1.3)
73. Decreasing: 10, 14 Increasing: 31, ∞ 2 y
5 21. (A) x intercepts: - 4, 0; y intercept: 0 (B) Vertex: 1 - 2, - 42 (C) Min-
imum: - 4 (D) Range: y Ú - 4 or 3 - 4, ∞ 2 (1.3) 22. Quadratic (1.3)
23. Linear (1.1) 24. None (1.1, 1.3) 25. Constant (1.1) 26. x = 8 (1.6)
27. x = 3 (1.6) 28. x = 3 (1.5) 29. x = - 1, 3 (1.5) 30. x = 0, 13 (1.5)
5 x 31. x = - 2 (1.6) 32. x = 12 (1.6) 33. x = 27 (1.6) 34. x = 13.3113 (1.6)
75. Increasing: 1 - 4, ∞ 2 77. Increasing: 10, ∞ 2 35. x = 158.7552 (1.6) 36. x = 0.0097 (1.6) 37. x = 1.4359 (1.6)
y y 38. x = 1.4650 (1.6) 39. x = 230.2609 (1.6) 40. x = 9.0065 (1.6)
10 10 41. x = 2.1081 (1.6) 42. (A) All real numbers except x = - 2 and 3
1 2
(B) x 6 5 (1.1) 43. Vertex form: 4 a x + b - 4; x intercepts: - 32 and 12 ;
10 x 2
10 x y intercept: - 3; vertex: 1 - 12 , - 42; minimum: - 4; range: y Ú - 4 or
3 - 4, ∞ 2 (1.3) 44. 10.99, 0.692; 1 - 0.79, - 1.542 (1.5, 1.6)
79. Because b0 = 1 for any permissible base b1b 7 0, b ≠ 12. y
45. (1.1) 46. y (1.1) 47. 6 (1.1)
81. x 7 1x 7 ln x for 1 6 x … 16 83. 4 yr 85. 9.87 yr; 9.80 yr 50 10
5
87. 7.51 yr 89. (A) 5,373 x
5 x
48. - 19 (1.1) 49. 10x - 4 (1.1) 50. 21 - 5x (1.1) 51. (A) - 1

(B) - 1 - 2h (C) - 2h (D) - 2 (1.1) 52. (A) - 5 (B) - 5 - 3h
(B) 7,220 (C) - 3h (D) - 3 (1.1) 53. The graph of function m is the graph of
y = x reflected in the x axis and shifted to the right 6 units. (1.2) 54. The
graph of function g is the graph of y = x3 vertically shrunk by a factor of 0.6
and shifted up 5 units. (1.2) 55. The graph of y = x2 is vertically expanded
by a factor of 2, reflected in the x axis, and shifted to the left 3 units. Equation:
y = - 21x + 32 2. (1.2) 56. f1x2 = 2 2x + 3 - 1 (1.2)
3
93. 168 bushels/acre 95. 912 yr f(x) 57. y = 0 (1.4) 58. y = (1.4) 59. None (1.4)
5 4
60. x = - 10, x = 10 (1.4) 61. x = - 2 (1.4)
62. True (1.3) 63. False (1.3) 64. False (1.3)
5 x
65. True (1.4) 66. True (1.5) 67. True (1.3)
A-6 Answers
68. y (1.2) 69. y (1.2) 89. (A) P1x2 = R1x2 - C1x2 = x150 - 1.25x2 - 1160 + 10x2
15 80
(B) P = 0 for x = 4.686 thou-
sand units (4,686 units) and for
x = 27.314 thousand units (27,314
40 x
50 x
units); P 6 0 for 1 … x 6 4.686
70. y = - 1x - 42 2 + 3 (1.2, 1.3) 71. f1x2 = - 0.41x - 42 2 + 7.6 or 27.314 6 x … 40; p 7 0 for
(A) x intercepts: - 0.4, 8.4; y intercept: 1.2 (B) Vertex: 14.0, 7.62 4.686 6 x 6 27.314. (C) Maximum
(C) Maximum: 7.6 (D) Range: y … 7.6 or 1 - ∞, 7.64 (1.3) profit is 160 thousand dollars ($160,000).
This occurs at an output of 16 thousand
72. (A) x intercepts: - .4, 8.4; y in-
units (16,000 units). At this output, the
tercept: 1.2 (B) Vertex:
wholesale price is p1162 = $30. (1.3)
(4.0, 7.6) (C) Maximum:
7.6 (D) Range: y … 7.6 or 90. (A) A1x2 = - 32 x2 + 420x (B) Domain: 0 … x … 280
1 - ∞, 7.6] (1.3) (C) (D) There are two solutions to the equation
A(x)
30,000 A1x2 = 25,000, one near 90 and another
near 190. (E) 86 ft; 194 ft (F) Maximum
73. log 10p = p and 10log 12 = 12; ln ep = p and eln12 = 12 (1.6) combined area is 29,400 ft2. This occurs for
74. x = 4 (1.6) 75. x = 2 (1.6) 76. x = 67 (1.6) 77. x = 300 (1.6) x = 140 ft and y = 105 ft. (1.3)
78. y = be - 7t (1.6) 79. If log1 x = y, then 1y = x; that is, 1 = x for 300 x
all positive real numbers x, which is not possible. (1.6) 80. The graph of
3
y = 2 x is vertically expanded by a factor of 2, reflected in the x axis, and 91. (A) 2,833 sets
3
shifted 1 unit left and 1 unit down. Equation: y = - 2 2x + 1 - 1. (1.2)
81. G1x2 = 0.41x + 22 2 - 8.1 (A) x int.: - 6.527, 2.527; y int.: - 6.5
(B) Vertex: 1 - 2, - 8.12 (C) Min.: - 8.1 (D) Range: y Ú - 8.1 or
3 - 8.1, ∞ 2 (1.3)
82. (A) x intercepts: - 7.2. 3.2;
y intercept: - 6.9 (B) Vertex: (B) 4,836 (C) Equilibrium price: $131.59; equi-
1 - 2, - 8.12 (C) Minimum: librium quantity: 3,587 cookware
- 8.1 (D) Range: sets (1.3)
y Ú - 8.1 or [ - 8.1,∞ 2 (1.3)
3 if 0 … x … 20
0.057x + 1.86 if 20 6 x … 200 92. (A) (B) 4976 (1.4)
83. (A) S1x2 = µ
0.0346x + 6.34 if 200 6 x … 1,000
0.0217x + 19.24 if x 7 1,000
(B) S(x) (1.2) 84. $5,321.95 (1.5)

50
85. $5,269.51 (1.5)
93. (A) N = 22t or N = 4t (B) 15 days (1.5) 94. k = 0.009 42; 489 ft (1.6)
95. (A) 6,134,000 (1.6)
1,500 x
86. 201 months 1 ≈16.7 years2 (1.5) 87. 9.38 yr (1.5)
88. (A) R(x) C(x) (B) R = C for x = 4.686 thou-
600 C
sand units (4,686 units) and for
x = 27.314 thousand units (27,314 96. 23.1 yr (1.5) 97. (A) $1,319 billion
units); R 6 C for 1 … x 6 4.686 (B) 2031 (1.5)
R
or 27.314 6 x … 40; R 7 C for
40 x
4.686 6 x 6 27.314. (C) Maxi-
mum revenue is 500 thousand dollars
($500,000). This occurs at an output
of 20 thousand units (20,000 units).
At this output, the wholesale price is Chapter 2
p1202 = $25. (1.3)
Exercises 2.1
1. $25.42 3. 64% 5. Slope = 120; y int. = 12,000
7. Slope = 50;
1
y int. = 2,000 9. 0.062 11. 13.7% 13. 0.0025 15. 0.84% 17. yr
1 5 6 2
19. yr 21. yr 23. yr 25. $42 27. $1,800 29. 0.12 or 12%
3 4 13
1
31. yr 33. $4,612.50 35. $875 37. 0.27 or 27% 39. 1 yr
2
Answers A-7
A - P
41. r = I> 1Pt2 43. P = A> 11 + rt2 45. t = 47. The graphs are 35. Period Amount Interest Balance
Pr
linear, all with y intercept $1,000; their slopes are 40, 80, and 120, respectively. 1 $1,000.00 $0.00 $1,000.00
49. $45 51. $30 53. $7,647.20 55. 8.1% 57. 18% 59. $1,604.40; 2 $1,000.00 $83.20 $2,083.20
20.88% 61. 4.298% 63. $992.38 65. $24.31 67. $27 69. 5.396% 3 $1,000.00 $173.32 $3,256.52
71. 6.986% 73. 12.085% 75. 109.895% 77. 87.158% 79. $118.94 4 $1,000.00 $270.94 $4,527.46
81. $1,445.89 83. $7.27 85. $824.85 87. 292% 89. 60% 5 $1,000.00 $376.69 $5,904.15
37. First year: $33.56; second year: $109.64; third year: $190.41
Exercises 2.2
39. $111,050.77 41. $1,308.75 43. (A) 1.650% (B) $528.89
1. P = 950 3. x = 17 5. i = 0.5 7. n = 4 9. $5,983.40 45. 33 months 47. 7.77% 49. 0.75% 51. After 11 quarterly
11. $4,245.07 13. $3,125.79 15. $1,733.04 17. 2.5 yr 19. 6.79%
21. 1.65% 23. 0.85% 25. 0.02% 27. 2.43% 29. 3.46% 31. 6.36%
Exercises 2.4
33. 8.76% 35. 2.19% 37. (A) $126.25; $26.25 (B) $126.90; $26.90
(C) $127.05; $27.05 39. (A) $5,524.71 (B) $6,104.48 41. $12,175.69 1. 511/256 3. 33,333,333/1,000,000 5. 547/729 7. i = 0.0045; n = 84
43. All three graphs are increasing, curve upward, and have the same y inter- 9. i = 0.02475; n = 40 11. i = 0.03375; n = 50 13. i = 0.0548; n = 9
cept; the greater the interest rate, the greater the increase. The amounts at the 15. PV = $3,458.41 17. PMT = $586.01 19. n = 29 21. i = 0.029
end of 8 years are $1,376.40, $1,892.46, and $2,599.27, respectively. 27. $60,304.25 29. $11,241.81; $1,358.19 31. $69.58; $839.84
45. 47. (A) $7,440.94 33. 31 months 35. 71 months 37. For 0% financing, the monthly payments
Period Interest Amount
(B) $5,536.76 should be $242.85, not $299. If a loan of $17,485 is amortized in 72 payments
0 $1,000.00 of $299, the rate is 7.11% compounded monthly. 39. The monthly payments
49. (A) $19,084.49
1 $97.50 $1,097.50 with 0% financing are $455. If you take the rebate, the monthly payments are
(B) $11,121.45
2 $107.01 $1,204.51 $434.24. You should choose the rebate. 41. $314.72; $17,319.68
51. (A) 3.97%
3 $117.44 $1,321.95 43. Unpaid
(B) 2.32%
4 $128.89 $1,450.84 53. (A) 5.28% Payment Balance Unpaid
5 $141.46 $1,592.29 (B) 5.27% Number Payment Interest Reduction Balance
6 $155.25 $1,747.54 2 0 $5,000.00
55. 11 yr
3 1 $706.29 $140.00 $566.29 4,433.71
1 2 706.29 124.14 582.15 3,851.56
57. 6 yr 59. n ≈ 12 61. (A) 7 yr (B) 6 yr 63. (A) 23.1 yr
4 3 706.29 107.84 598.45 3,253.11
(B) 11.6 yr 65. $65,068.44 67. $282,222.44 69. $19.78 per ft2 per mo 4 706.29 91.09 615.20 2,637.91
71. (A) In 2026, 250 years after the signing, it would be worth $175,814.55. 5 706.29 73.86 632.43 2,005.48
(B) If interest were compounded monthly, daily, or continuously, it would be 6 706.29 56.15 650.14 1,355.34
worth $179,119.92, $180,748.53, or $180,804.24, respectively. 7 706.29 37.95 668.34 687.00
(C) 73. 9.66% 75. 2 yr, 10 mo 8 706.24 19.24 687.00 0.00
77. $163,295.21 79. 3,615 days;
Totals $5,650.27 $650.27 $5,000.00
8.453 yr
45. First year: $466.05; second year: $294.93; third year: $107.82
47. £15,668.64; £1,083.36 49. $143.85/mo; $904.80 51. Monthly pay-
ment: $908.99 (A) $125,862 (B) $81,507 (C) $46,905 53. (A) Monthly
81. 83. 14 quarters payment: $1,015.68; interest: $114,763 (B) 197 months; interest saved:
Years Exact Rate Rule of 72
85. To maximize $23,499 55. (A) 157 (B) 243 (C) The withdrawals continue forever.
6 12.2 12.0
earnings, choose 10% 57. (A) Monthly withdrawals: $1,229.66; total interest: $185,338.80
7 10.4 10.3
simple interest for (B) Monthly deposits: $162.65 59. $65,584 61. $34,692 63. All three
8 9.1 9.0 graphs are decreasing, curve downward, and have the same x intercept; the
investments lasting fewer
9 8.0 8.0 unpaid balances are always in the ratio 2:3:4. The monthly payments are
than 11 years and
10 7.2 7.2 7% compound interest $402.31, $603.47, and $804.62, with total interest amounting to $94,831.60,
11 6.5 6.5 otherwise. 87. 3.33% $142,249.20, and $189,663.20, respectively. 65. 14.45% 67. 10.21%
12 5.9 6.0
Chapter 2 Review Exercises
89. 7.02% 91. $15,843.80 93. 4.53% 95. 13.44% 97. 17.62%
1. A = $212 (2.1) 2. P = $869.57 (2.1) 3. t = 3.543 yr, or 43 mo (2.1)
4. r = 5% (2.1) 5. A = $1,578.12 (2.2) 6. P = $4,140.73 (2.2)
7. A = $6,064.18 (2.2) 8. P = $25,601.80 (2.2) 9. PMT = $117.95 (2.3)
Exercises 2.3 10. FV = $99,897.40 (2.3) 11. PV = $39,536.26 (2.4)
1. 1,023 3. 3,000 5. 71,744,530 7. i = 0.02; n = 80 9. i = 0.0375; 12. PMT = $594.08 (2.4) 13. n ≈ 16 (2.2) 14. n ≈ 41 (2.3)
n = 24 11. i = 0.0065; n = 84 13. i = 0.0625; n = 15 15. $4,570.00; $570.00 (2.1) 16. $19,654 (2.2) 17. $12,944.67 (2.2)
15. FV = $13,435.19 17. PMT = $310.62 19. n = 17 21. i = 0.09 18. (A)
Period Interest Amount
FV
ln a 1 + i b 0 $400.00
PMT
25. n = 27. Value: $84,895.40; interest: $24,895.40 1 $21.60 $421.60
ln 11 + i2
2 $22.77 $444.37
29. $20,931.01 31. $176.25 33. $763.39
3 $24.00 $468.36
4 $25.29 $493.65 (2.2)
A-8 Answers
(B) Period Interest Payment Balance 61. 16 63. (A) 120, - 242 (B) 1 - 4, 62
2,
3
1 $100.00 $100.00 y 8x 6y 24 (C) No solution 65. (A) Supply: 143
10 T-shirts; demand: 647 T-shirts
2 $5.40 $100.00 $205.40 2x 3y 12
10 (B) Supply: 857 T-shirts; demand: 353
3 $11.09 $100.00 $316.49 x ; no
T-shirts (C) Equilibrium price = $6.50;
4 $17.09 $100.00 $433.58 (2.3) 4x 3y 24 equilibrium quantity = 500 T-shirts
4
19. To maximize earnings, choose 13% simple interest for investments lasting 4,
3
less than 9 years and 9% compound interest for investments lasting 9 years or p 67. (A) p = 1.5x + 1.95
(D) Equilibrium
more. (2.2) 20. $164,402 (2.2) 21. 7.83% (2.2) 22. 9% compounded 15 point (B) p = - 1.5x + 7.8 (C) Equilibrium
Supply
Price ($)
quarterly, since its effective rate is 9.31%, while the effective rate of 9.25% (5, 6.5) curve price: $4.875; equilibrium quantity:
compounded annually is 9.25% (2.2) 23. $25,861.65; $6,661.65 (2.3) 1.95 billion bushels
Demand
24. 288% (2.1) 25. $1,725.56 (2.1) 26. $29,354 (2.2) 27. $18,021 (2.2) curve (D) Equilibrium
point Supply
28. 15% (2.1) 29. The monthly payments with 0% financing are $450. If 10 q p curve
Quantity (hundreds)
you take the rebate, the monthly payments are $426.66. You should choose 12
the rebate. (2.4) 30. (A) 6.43% (B) 6.45% (2.2) 31. 9 quarters or 2 yr,
3 mo (2.2) 32. 139 mo; 93 mo (2.2) 33. (A) $571,499 (B) $1,973,277 (2.3) 6
(1.95, 4.875)
34. 10.45% (2.2) 35. (A) 174% (B) 65.71% (2.1) 36. $725.89 (2.3) Demand
curve
37. (A) $140,945.57 (B) $789.65 (C) $136,828 (2.3, 2.4) 38. $102.99; 0 3 6 x
$943.52 (2.4) 39. $576.48 (2.3) 40. 3,374 days; 10 yr (2.2) 41. $175.28; 69. (A) 120 mowers 71. (A) C = 24,000 + 7.45x; R = 19.95x
$2,516.80 (2.4) 42. $13,418.78 (2.2) 43. 5 yr, 10 mo (2.3) 44. 18 yr (2.4) (B) y Revenue line (B) 1,920
$400,000 Profit Cost
45. 28.8% (2.1) line
(C) Revenue line
y
46. $60,000
Unpaid Loss (120, 216,000)
Profit Cost
Payment Balance Unpaid Break-even point Loss
line
Number Payment Interest Reduction Balance 200 x (1,920, 38,304)

Break-even point
0 $1,000.00
3,000 x
1 $265.82 $25.00 $240.82 759.18
2 265.82 18.98 246.84 512.34 73. Base price = :10.56; surcharge = :1.56>kg 75. 5,720 lb robust
3 265.82 12.81 253.01 259.33 blend; 6,160 lb mild blend 77. Mix A: 80 g; mix B: 60 g 79. Operate the
Mexico plant for 75 hours and the Taiwan plant for 50 hours. 81. (A) a = 196,
4 265.81 6.48 259.33 0.00
b = - 16 (B) 196 ft (C) 3.5 sec 83. 40 sec, 24 sec, 120 mi
Totals $1,063.27 $63.27 $1,000.00 p
85. (A) (B) d = 141 cm (approx.) (C) Vacillate
200
47. 28 months (2.3) 48. $55,347.48; $185,830.24 (2.3) 49. 2.47% (2.2) Avoidance
Pull (g)
50. 6.33% (2.1) 51. 44 deposits (2.3) 52. (A) $1,189.52 (B) $72,963.07
(C) $7,237.31 (2.4) 53. The certificate would be worth $53,394.30 when the Approach
360th payment is made. By reducing the principal, the loan would be paid off (141, 42)
200 d
in 252 months. If the monthly payment were then invested at 7% compounded Distance (cm)
monthly, it would be worth $67,234.20 at the time of the 360th payment.
(2.2, 2.3, 2.4) 54. The lower rate would save $12,247.20 in interest pay- Exercises 3.2
ments. (2.4) 55. $3,807.59 (2.2) 56. 5.79% (2.2) 57. $4,844.96 (2.1)
1. 6; 3 3. 3 * 3; 2 * 1 5. D 7. B 9. 2, 1 11. -4, 9, 0 13. - 1
58. $6,697.11 (2.4) 59. 7.24% (2.2) 60. (A) $398,807 (B) $374,204 (2.3)
3 5 3 5 8 1 4 1 4 15
61. $15,577.64 (3.2) 62. (A) 30 yr: $569.26; 15 yr: $749.82 (B) 30 yr: 15. c d; c ` d 17. c d; c ` d
2 -4 2 -4 -7 6 0 6 0 18
$69,707.99; 15 yr: $37,260.74 (2.4) 63. $20,516 (2.4) 64. 33.52% (2.4)
19. 2x1 + 5x2 = 7 21. 4x1 = - 10
65. (A) 10.74% (B) 15 yr: 40 yr (2.3)
x1 + 4x2 = 9 8x2 = 40
1 -3 5 2 -4 6 3 - 7 11
23. c ` d 25. c ` d 27. c ` d
2 -4 6 2 - 6 10 1 -3 5
Chapter 3 29. c
-1
1 -3
2 -3
`
5
d 31. c
1 -1 1
` d 33. c
1 -3 5
-1
1 -3
5 -9
`
5
d
Exercises 3.1 35. 13 R2 S R2 37. 6R1 + R2 S R2 39. 13 R2 + R1 S R1 41. R1 4 R2
1. 10, 72 3. 124, 02 5. 15, - 182 7. y - 7 = - 61x - 22 9. (B); no 43. 516, 626 x2 45. 51 23 t - 1, t 2 t is any real
solution 11. (A); x = - 3, y = 1 13. x = 2, y = 4 15. No solution number 6
10
(parallel lines) 17. x = 4, y = 5 19. x = 1, y = 4 21. u = 2, v = - 3 x2
23. m = 8, n = 6 25. x = 1, y = 1 27. No solution (inconsistent)
10
29. Infinitely many solutions (dependent) 31. m = 12, n = 1110 33. x = - 1, x1
10 10
y = 2 35. x = 7, y = 3 37. x = 76 , y = - 89 39. x = 0, y = 0
41. x = 14, y = 5 43. Price tends to come down. 49. 11.125, 0.1252 x1
10 10
51. No solution (parallel lines) 53. 14.176, - 1.2352 55. 1 - 3.310, - 2.2412 10
57. y 59. y
10
( 4, 1) 8 (2, 4) 10
x 2y 8
x 2y 2 (6, 4) ; no ( 2, 3) ; yes
10 x 10 x
x 2y 6 x y 1
2x y 8 3x y 3
Answers A-9
47. x1 = 3, x2 = 2; each pair of lines has the same intersection point. 79. x1 = federal income tax
x2 x2 x2 x2
x2 = state income tax
6 6 6 6
(3, 2) (3, 2) (3, 2) x3 = local income tax
(3, 2)
x1 + 0.5x2 + 0.5x3 = 3,825,000
5 x1 5 x1 5 x1 5 x1
0.2x1 + x2 + 0.2x3 = 1,530,000
0.1x1 + 0.1x2 + x3 = 765,000
x1 x2 5 x1 x2 5 x1 x2 5 x1 3
x1 x2 1 2x 2 4 x2 2 x2 2
Tax liability is 57.65%.
81. x1 = taxable income of company A
49. x1 = - 4, x2 = 6 51. No solution 53. x1 = 2t + 15, x2 = t for any
x2 = taxable income of company B
real number t 55. x1 = 3, x2 = 1 57. x1 = 3, x2 = - 2 59. x1 = 2, x2 = 4
x3 = taxable income of company C
61. No solution 63. x1 = 1, x2 = 4 65. Infinitely many solutions: x2 = s,
x4 = taxable income of company D
x1 = 2s - 3 for any real number s 67. Infinitely many solutions; x2 = s,
x1 - 0.08x2 - 0.03x3 - 0.07x4 = 2.272
x1 = 12 s + 12 for any real number s 69. x1 = - 1, x2 = 3 71. No solution
- 0.12x1 + x2 - 0.11x3 - 0.13x4 = 2.106
73. Infinitely many solutions: x2 = t, x1 = 43 t + 3 for any real number t
- 0.11x1 - 0.09x2 + x3 - 0.08x4 = 2.736
75. x1 = 2, x2 = - 1 77. x1 = 2, x2 = - 1 79. x1 = 1.1, x2 = 0.3
- 0.06x1 - 0.02x2 - 0.14x3 + x4 = 3.168
81. x1 = - 23.125, x2 = 7.8125 83. x1 = 3.225, x2 = - 6.9375
Taxable incomes are $2,927,000 for company A, $3,372,000 for company
B, $3,675,000 for company C, and $3,926,000 for company D.
Exercises 3.3 83. (A) x1 = no. of ounces of food A (B) 30x1 + 10x2 = 340
1 0 6 2 5. x1 - 3x2 = 4 x2 = no. of ounces of food B 10x1 + 10x2 = 180
1 2 3 12
1. c ` d
3. £ 0 1 - 1 † 5 § 3x1 + 2x2 = 5 x3 = no. of ounces of food C 10x1 + 30x2 = 220
1 7 - 5 15
1 3 0 7 - x1 + 6x2 = 3 30x1 + 10x2 + 20x3 = 340 There is no combination that
7. 5x1 - 2x2 + 8x4 = 4 9. Reduced form 11. Not reduced form; 10x1 + 10x2 + 20x3 = 180 will meet all the requirements.
R1 4 R2 13. Not reduced form; 14 R1 S R1 15. Not reduced form; 10x1 + 30x2 + 20x3 = 220
- 7 R2 + R3 S R3 17. Not reduced form; - 12 R2 S R2 19. x1 = - 2, 8 oz of food A, 2 oz of food B,
x2 = 3, x3 = 0 21. x1 = 2t + 3, x2 = - t - 5, x3 = t for any real number t and 4 oz of food C
23. No solution 25. x1 = 3t + 5, x2 = - 2t - 7, x3 = t for any real number t (C) 30x1 + 10x2 + 20x3 = 340
27. x1 = 2s + 3t - 5, x2 = s, x3 = - 3t + 2, x4 = t for any real numbers s 10x1 + 10x2 + 20x3 = 180
and t 29. 19 31. 21, 25, 27 33. False 35. True 37. False 8 oz of food A, 110 - 2t2 oz
1 0 0 -5 of food B, and t oz of food C
1 0 -7 1 0 - 1 23
39. c ` d 41. c ` d 43. £ 0 1 0 † 4 § where 0 … t … 5
0 1 3 0 1 2 -7
0 0 1 -2 85. x1 = no. of barrels of mix A 30x1 + 30x2 + 30x3 + 60x4 = 900
1 0 2 1 47. x1 = - 2, x2 = 3, x3 = 1 49. x1 = 0, x2 = no. of barrels of mix B 50x1 + 75x2 + 25x3 + 25x4 = 750
45. £ 0 1 - 2 † - 1 § x2 = - 2, x3 = 2 51. x1 = 2t + 3, x2 = t - 2, x3 = no. of barrels of mix C 30x1 + 20x2 + 20x3 + 50x4 = 700
0 0 0 0 x3 = t for any real number t 53. x1 = 1, x2 = 2 x4 = no. of barrels of mix D
55. No solution 57. x1 = t - 1, x2 = 2t + 2, x3 = t for any real number t 110 - t2 barrels of mix A, 1t - 52 barrels of mix B, 125 - 2t2 barrels of
59. x1 = - 2s + t + 1, x2 = s, x3 = t for any real numbers s and t 61. No mix C, and t barrels of mix D, where t is an integer satisfying 5 … t … 10
solution 63. (A) Dependent system with two parameters and an infinite 87. 0 barrels of mix A, 5 barrels of mix B, 5 barrels of mix C, and 10 barrels of mix D
number of solutions (B) Dependent system with one parameter and an 89. y = 0.0025x2 + 0.2x + 34,55 million 91. y = 0.004x2 + 0.06x + 77.6;
infinite number of solutions (C) Independent system with a unique solution 1995–2000: 79.4 years; 2000–2005: 80.4 years
(D) Impossible 65. x1 = 2s - 3t + 3, x2 = s + 2t + 2, x3 = s, x4 = t 93.
for s and t any real numbers 67. x1 = - 0.5, x2 = 0.2, x3 = 0.3, x4 = - 0.4
69. x1 = 2s - 1.5t + 1, x2 = s, x3 = - t + 1.5, x4 = 0.5t - 0.5, x5 = t
for any real numbers s and t 71. a = 2, b = - 4, c = - 7
73. (A) x1 = no. of one-person boats (B) 0.5x1 + x2 + 1.5x3 = 380
x2 = no. of two-person boats 0.6x1 + 0.9x2 + 1.2x3 = 330
x3 = no. of four-person boats 1t - 802 one-person boats,
0.5x1 + x2 + 1.5x3 = 380 1420 - 2t2 two-person 95. x1 = no. of hours for company A 97. (A) Old Street and Baker Street:
0.6x1 + 0.9x2 + 1.2x3 = 330 boats, and t four-person boats, x2 = no. of hours for company B x1 + x4 = 1,200; Bridge Street and
0.2x1 + 0.3x2 + 0.5x3 = 120 where t is an integer satisfying 30x1 + 20x2 = 600 Baker Street: x1 + x2 = 1,000;
20 one-person boats, 220 two-person 80 … t … 210 10x1 + 20x2 = 400 Bridge Street and Market Street:
boats, and 100 four-person boats Company A: 10 hr; x2 + x3 = 900; Old Street and
(C) 0.5x1 + x2 = 380 company B: 15 hr Market Street: x3 + x4 = 1,100
0.6x1 + 0.9x2 = 330 (B) x1 = 1,200 - t, x2 = t - 200, x3 = 1,100 - t, and x4 = t, where
0.2x1 + 0.3x2 = 120 200 … t … 1,100 (C) 1,100; 200 (D) Baker Street: 600; Bridge St.: 400;
There is no production schedule that will Market Street: 500
use all the labor-hours in all departments.
Exercises 3.4
75. x1 = no. of 8,000-gal tank cars 77. The minimum monthly
x2 = no. of 16,000-gal tank cars cost is $24,100 when 2 4
1. 34 154 3. c d 5. Not defined 7. 321 - 35 63 284
x3 = no. of 24,000-gal tank cars 716,000-gallon and 17 -2 6
x1 + x2 + x3 = 24 24,000-gallon tank cars are 5 2 4 1 -5 5 10 5 10
9. c d 11. c d 13. c d 15. c d 17. c d
8,000x1 + 16,000x2 + 24,000x3 = 520,000 leased. -3 1 -5 -2 -4 15 20 15 20
1t - 172 8,000-gal tank cars, 7 9 0 3 - 15 10
19. c d 21. c d 23. 3 - 234 25. c d 27. [1]
141 - 2t2 16,000-gal tank cars, and t 24,000-gal 0 0 0 7 12 -8
tank cars, where t = 17, 18, 19, or 20
A-10 Answers
-2 0 2 -1 6 71. M -1 exists if and only if all the elements on the main diagonal are nonzero.
- 12 12 18 73. A-1 = A; A2 = I 75. A-1 = A; A2 = I 77. 41 50 28
29. £ - 1 0 1§ 31. c d33. £ 13 - 3 §
20 - 18 -6 35 37 55 22 31 47 60 24 36 49 71 39 54 21 22
-3 0 3 - 1 11
- 1.3 - 0.7 79. PRIDE AND PREJUDICE
11 2 6 4 - 66 69 39 81. 37 47 10 58 103 67 47 123 121 75 53 142 58 68
35. c d 37. c d 39. £ - 0.2 - 0.5 § 41. c d
4 27 0 -3 92 - 18 - 36 23 91 90 74 38 117 83 59 39 103 113 97 45 147 76
0.1 1.1
- 26 - 15 - 25 57 38 95 83. RAWHIDE TO WALTER REED 85. 30 28 58 15
- 18 48 38 13 19 26 12 30 39 56 48 43 40 9 30 29 12 33
43. Not defined 45. c d 47. £ - 4 - 18 4§
54 - 34 87. DOUBLE DOUBLE TOIL AND TROUBLE
2 43 - 19
0 0 a2 + ab a2 + ab 0 0
49. AB = c d , BA = c 2 d 51. c d Exercises 3.6
0 0 - a - ab - a2 - ab 0 0
0.25 0.75 1. - 3>5 3. - 7>4 5. 9>8 7. 1>10 9. 3x1 + x2 = 5 2x1 - x2 = - 4
53. Bn approaches c d ; ABn approaches 30.25 0.754 11. - 3x1 + x2 = 3
0.25 0.75
55. a = - 1, b = 1, c = 3, d = - 5 57. a = 3, b = 4, c = 1, d = 2 2x1 + x3 = - 4
59. False 61. True 63. (A) True (B) True (C) True - x1 + 3x2 - 2x3 = 2
65. Guitar Banjo
13. 3 - 4 x1 1
$51.50 $40.50 Materials c dc d = c d
c d 2 1 x2 5
$87.00 $120.00 Labor
67. Basic AM/FM Cruise 1 -3 2 x1 -3 17. x1 = - 8, x2 = 2
car Air radio control 15. £ - 2 3 0 § £ x2 § = £ 1 § 19. x1 = - 29, x2 = 52
Model A $2,937 $459 $200 $118 1 1 4 x3 -2 21. x1 = 3, x2 = - 2
Model B £ $2,864 $201 $88 $52 §
23. x1 = 11, x2 = 4 25. x1 = 3, x2 = 2 27. No solution
Model C $2,171 $417 $177 $101
29. x1 = 2, x2 = 5 31. (A) x1 = - 3, x2 = 2 (B) x1 = - 1, x2 = 2
69. (A) €29.21 (B) €14.54 (C) MN gives the labor costs at each plant.
(C) x1 = - 8, x2 = 3 33. (A) x1 = 5, x2 = - 2 (B) x1 = - 35, x2 = 16
(D) IT PL
(C) x1 = 20, x2 = - 9 35. (A) x1 = 1, x2 = 0, x3 = 0
€18.68 €14.54 Small shelving unit
(B) x1 = - 7, x2 = - 2, x3 = 3 (C) x1 = 17, x2 = 5, x3 = - 7
MN = £ €29.21 €22.76 § Medium shelving unit
37. (A) x1 = 8, x2 = - 6, x3 = - 2 (B) x1 = - 6, x2 = 6, x3 = 2
€40.03 €31.24 Large shelving unit
(C) x1 = 20, x2 = - 16, x3 = - 10 39. X = A-1B 41. X = BA-1
71. (A) 70 g (B) 30 g (C) MN gives the amount (in grams) of protein,
43. X = A-1BA 45. No solution 47. No solution
carbohydrate, and fat in 20 oz of each mix.
49. x1 = 13t + 3, x2 = 8t + 1, x3 = t, for any real number t
(D) Mix X Mix Y Mix Z
51. X = 1A - B2 -1C 53. X = 1A + I2 -1C
70 60 50 Protein
55. X = 1A + B2 -1 1C + D2 57. x1 = 3, x2 = 8
MN = £ 380 360 340 § Carbohydrate
59. x1 = 10.2, x2 = 4.4, x3 = 7.3
50 40 30 Fat
61. x1 = 3.1, x2 = 4.3, x3 = - 2.7, x4 = 8
73. (A) £10,830 (B) £14,080
63. x1 = no. of $25 tickets
Cost per borough
£10,830 Greenwich x2 = no. of $35 tickets
(C)NM = c d x1 + x2 = 10,000 seats
£14,080 Bromley
The entries are the total cost per borough. 25x1 + 35x2 = k Return
(A) Concert 1: 7,500 $25 tickets, 2,500 $35 tickets
Concert 2: 5,000 $25 tickets, 5,000 $35 tickets
Concert 3: 2,500 $25 tickets, 7,500 $35 tickets
Exercises 3.5
(B) No (C) 250,000 + 10t, 0 … t … 10,000
1. (A) - 4; 1>4 (B) 3; - 1>3 3. (A) - 2>3; 3>2
(C) 0; not defined 65. x1 = no. of hours plant A operates
2 -3 2 0 x2 = no. of hours plant B operates
(B) 1>7; - 7 (C) - 1.6; 0.625 5. No 7. No 9. (A) c d (B) c d
0 0 4 0 10x1 + 8x2 = k1 no. of car frames produced
0 0 0 -3 2 -3 2 -3 5x1 + 8x2 = k2 no. of truck frames produced
11. (A) c d (B) c d 13. (A) c d (B) c d
4 5 0 5 4 5 4 5 Order 1: 280 hr at plant A and 25 hr at plant B
-2 1 3 -2 1 3 19. Yes 21. No 23. Yes Order 2: 160 hr at plant A and 150 hr at plant B
15. £ 2 4 - 2 § 17. £ 2 4 - 2 § 25. Yes 27. No Order 3: 80 hr at plant A and 225 hr at plant B
5 1 0 5 1 0 67. x1 = president’s bonus
-1 0 3 -2 9 -2 x2 = executive vice-president’s bonus
39. c d 41. c d 43. c d
-3 1 -1 1 -4 1 x3 = associate vice-president’s bonus
- 5 - 12 3 6 -2 -1 -2 -3 x4 = assistant vice-president’s bonus
49. c d
45. £ - 2 -4 1 § 47. £ - 5 2 1§ 3 4 x1 + 0.03x2 + 0.03x3 + 0.03x4 = 60,000
2 5 -1 -3 1 1 0.025x1 + x2 + 0.025x3 + 0.025x4 = 50,000
1.5 - 0.5 0.35 0.01 0.02x1 + 0.02x2 + x3 + 0.02x4 = 40,000
51. Does not exist 53. c d 55. c d
-2 1 0.05 0.03 0.015x1 + 0.015x2 + 0.015x3 + x4 = 30,000
1 1 1 2 2 President: $56,600; executive vice-president: $47,000; associate
0 0
57. c 3 1d 59. c 2 d
61. £ - 2 - 3 -4 § vice-president: $37,400; assistant vice-president: $27,900
0 0 2
3
-1 -2 -1 69. (A) x1 = no. of ounces of mix A
-3 -2 1.5 - 1.75 - 0.375 0.5 x2 = no. of ounces of mix B
63. Does not exist 65. £ 4 3 -2 § 67. £ - 5.5 - 1.25 1 § 0.20x1 + 0.14x2 = k1 Protein
3 2 - 1.25 0.5 0.25 0
Answers A-11
0.04x1 + 0.03x2 = k2 Fat - 0.9 - 0.1 5

Diet 1: 50 oz mix A and 500 oz mix B 33. £ 0.8 0.2 - 4 § (3.5) 34. x1 = 1,400, x2 = 3,200, x3 = 2,400 (3.6)
Diet 2: 450 oz mix A and 0 oz mix B 0.1 - 0.1 0
Diet 3: 150 oz mix A and 500 oz mix B (B) No 35. x1 = 1,400, x2 = 3,200, x3 = 2,400 (3.3)
1.3 0.4 0.7 81
Exercises 3.7 36. (I-M) -1 = £ 0.2 1.6 0.3 § ; X = £ 49 § (3.7)
1. - 3 3. 100 5. 4 7. 8 9. 40 c from A; 20c from E 0.1 0.8 1.4 62
0.6 - 0.2 1.8 0.4 x 16.4 37. (A) Unique solution (B) Either no solution or an infinite no. of solutions (3.6)
11. c d; c d 13. X = c 1 d = c d
- 0.2 0.9 0.4 1.2 x2 9.2 38. (A) Unique solution (B) No solution (C) Infinite no. of solutions (3.3)
0.8 - 0.3 - 0.2 39. (B) is the only correct solution. (3.6) 40. (A) C = 253,000 + 35.50x;
15. 30c from A; 20 c from B; 30 c from E 17. £ - 0.1 0.8 - 0.1 § R = 199.99x (B) x = 1,538 computers; R = C = $307,602.10 (C) Profit
- 0.1 - 0.3 0.9 occurs if x 7 1,538; loss occurs if x 6 1,538. (3.1)
19. Agriculture: $31.3 billion; building: $27 billion; energy: $34.7 billion y 41. x1 = no. of tons Port Hedland ore
1.75 1 78.75 $100,000 R
21. I - M is singular; X does not exist 23. c d; c d x2 = no. of tons of Jack Hills ore
1 2 95 $800,000
$600,000 C
0.02x1 + 0.03x2 = 6
1.58 0.24 0.58 38.6 27. (A) Agriculture: $80 million; manu-
$400,000 0.04x1 + 0.02x2 = 8
25. £ 0.4 1.2 0.4 § ; £ 18 § facturing: $64 million. (B) The final $200,000 (1,538, 307,602.10)
Break-even point x1 = 150 tons of Port Hedland ore
0.22 0.16 1.22 17.4 demand for agriculture increases to $54
0 2,000 5,000 x x2 = 100 tons of Jack Hills ore (3.3)
0.25 0.1 million and the final demand for manu-
29. c d x1 - 25 37.5 6 150
0.25 0.3 facturing decreases to $38 million. 42. (A) c d = c dc d = c d
31. The total output of the energy sector should be 75% of the total output of x2 50 - 25 8 100
the mining sector. x1 = 150 tons of Port Hedland ore
33. Each element should be between 0 and 1, inclusive. x2 = 100 tons of Jack Hills ore
35. Coal: $28 billion; steel: $26 billion x - 25 37.5 7.5 75
(B) c 1 d = c dc d = c d
37. Agriculture: $165 million; tourism: $164 million x2 50 - 25 7 200
x1 = 75 tons of Port Hedland ore
39. Agriculture: $40.1 billion; manufacturing: $29.4 billion; energy: $34.4
x2 = 200 tons of Jack Hills ore (3.6)
billion
43. (A) x1 = no. of 3,000-ft3 hoppers
41. Year 1: agriculture: $65 billion; energy: $83 billion; labor: $71 billion;
x2 = no. of 4,500-ft3 hoppers
manufacturing: $88 billion
x3 = no. of 6,000-ft3 hoppers
Year 2: agriculture: $81 billion; energy: $97 billion; labor: $83 billion; manu-
x1 + x2 + x3 = 20
facturing: $99 billion
3,000x1 + 4,500x2 + 6,000x3 = 108,000
Year 3: agriculture: $117 billion; energy: $124 billion; labor: $106 billion;
x1 = (t - 12) 3,000-ft3 hoppers
manufacturing: $120 billion
x2 = 132 - 2t2 4,500-ft3 hoppers
Chapter 3 Review Exercises x3 = t 6,000-ft3 hoppers
where t = 12, 13, 14, 15, or 16
1. x = 4, y = 4 (3.1) 2. x = 4, y = 4 (3.1) 3. (A) Not in reduced form;
(B) The minimum monthly cost is $5,700 when 8 4,500-ft3 and
R1 4 R2 (B) Not in reduced form; 13 R2 4 R2 (C) Reduced form (D) Not
12 6,000-ft3 hoppers are leased. (3.3)
in reduced form; 1 - 12R2 + R1 S R1 (3.3) 4. (A) 2 * 5, 3 * 2
44. (A) Elements in MN give the cost of materials for each alloy from each
(B) a24 = 3, a15 = 2, b31 = -1, b22 = 4 (C) AB is not defined; BA is defined (3.2,
supplier.
8 7
3.4) 5. (A) x1 = 8, x2 = 2 (B) x1 = - 15.5, x2 = 23.5 (3.6) 6. c d (3.4) (B) Supplier A Supplier B
8 5
$7,620 $7,530 Alloy 1
7 8 22 MN = c d
7. Not defined (3.4) 8. c d (3.4) 9. c d (3.4) 10. Not defined (3.4) $13,880 $13,930 Alloy 2
10 4 8
(C) Supplier A Supplier B
5 2 3
11. c d (3.4) 12. c d (3.4) 13. 322 164 (3.4) 14. Not defined (3.4) 3114MN = 3 $21,500 $21,4604
5 4 6
Total material costs (3.4)
-2 3
15. c d (3.5) 16. m1 = - 3, m2 = 3 (3.1) 17. x1 = 9, x2 = - 11 (3.2) 45. (A) $6.35 (B) Elements in MN give the total labor costs for each
3 -4
calculator at each plant.
18. x1 = 9,x2 = - 11; x1 = 16, x2 = - 19; x1 = - 2, x2 = 4 (3.6)
(C) CA TX
10 - 8 -2 8
19. Not defined (3.4) 20. c d (3.4) 21. c d (3.4) $3.65 $3.00 Model A
4 6 8 6 MN = c d (3.4)
$6.35 $5.20 Model B
-2 -1 -3 10 - 5 1 46. x1 = amount invested at 5% x2 = amount invested at 10%
22. £ 4 2 6 § (3.4) 23. 394 (3.4) 24. £ - 1 - 4 - 5 § (3.4) x1 + x2 = 5,000 0.05x1 + 0.1x2 = 400 $2,000 at 5%, $3,000 at 10% (3.3)
6 3 9 1 -7 -2 47. $2,000 at 5% and $3,000 at 10% (3.6) 48. No to both. The annual yield
- 52 2 - 12 26. (A) x1 = 2, x2 = 1, x3 = - 1 must be between $250 and $500 inclusive. (3.6)
25. £ 1 - 1 1 § (3.5) (B) x1 = - 5t - 12, x2 = 3t + 7, x3 = t for t any 49. x1 = no. of $8 tickets
1
2 0 - 1
2
real number (C) x1 = - 2t + 5, x2 = no. of $12 tickets
x2 = t + 3, x3 = t for t any real number (3.3) x3 = no. of $20 tickets
27. x1 = 2, x2 = 1, x3 = - 1; x1 = 1, x2 = - 2, x3 = 1; x1 = - 1, x1 + x2 + x3 = 25,000
x2 = 2, x3 = - 2 (3.6) 28. The system has an infinite no. of solutions for 8x1 + 12x2 + 20x3 = k1 Return required
k = 3 and a unique solution for any other value of k. (3.3) x1 - x3 = 0
1.4 0.3 48 0.3 0.4 Concert 1: 5000 $8 tickets, 15,000 $12 tickets, and 5,000 $20 tickets
29. 1I - M2 -1 = c d ; X = c d (3.7) 30. c d (3.7)
0.8 1.6 56 0.15 0.2 Concert 2: 7,500 $8 tickets, 10,000 $12 tickets, and 7,500 $20 tickets
31. I - M is singular; X does not exist. (3.5) 32. x = 3.46, y = 1.69 (3.1) Concert 3: 10,000 $8 tickets, 5,000 $12 tickets, and 10,000 $20 tickets (3.6)
A-12 Answers
50. x1 + x2 + x3 = 25,000 57. Let x = no. of lbs of brand A.

8x1 + 12x2 + 20x3 = k1 Return required Let y = no. of lbs of brand B.
Concert 1: 12t - 5,0002 $8 tickets, 130,000 - 3t2 $12 tickets, and t (A) 0.26x + 0.16y Ú 120, (B) 0.03x + 0.08y … 28,
$20 tickets, where t is an integer satisfying 2,500 … t … 10,000 x Ú 0, y Ú 0 x Ú 0, y Ú 0
y
Concert 2: 12t - 7,5002 $8 tickets, 132,500 - 3t2 $12 tickets, and t y
1500 600
$20 tickets, where t is an integer satisfying 3,750 … t … 10,833
Concert 3: 12t - 10,0002 $8 tickets, 135,000 - 3t2 $12 tickets, and t
$20 tickets, where t is an integer satisfying 5,000 … t … 11,666 (3.3)
51. (A) Wheat: 825 metric tons; Oil: 1425 metric tons
1500 x 1500 x
(B) Wheat: 1372 metric tons; Oil: 2850 metric tons (3.7)
52. BEWARE THE IDES OF MARCH (3.5) 59. Let x = no. of kilograms of commercial bronze. y
10
53. (A) 1st and Elm: x1 + x4 = 1,300 Let y = no. of kilograms of architechtural bronze.
2nd and Elm: x1 - x2 = 400 0.90x + 0.57y Ú 0.751x + y2, x Ú 0, y Ú 0
2nd and Oak: x2 + x3 = 700
1st and Oak: x3 - x4 = - 200
(B) x1 = 1,300 - t , x2 = 900 - t, x3 = t - 200, x4 = t, 10 x
61. Let x = no. of weeks Plant A is operated. y
where 200 … t … 900 90
Let y = no. of weeks Plant B is operated.
(C) 900; 200
10x + 8y Ú 400, x Ú 0, y Ú 0
(D) Elm St.: 800; 2nd St.: 400; Oak St.: 300 (3.3)
90 x
Chapter 4 63. Let x = no. of radio spots. y

25
Let y = no. of television spots.
Exercises 4.1 200x + 800y … 10,000, x Ú 0, y Ú 0
1. No 3. Yes 5. No 7. No 9. y 11. y
5 5
80 x
5 x 5 x 65. Let x = no. of regular mattresses cut per day. y

800
Let y = no. of king mattresses cut per day
5x + 6y … 3,000, x Ú 0, y Ú 0
13. y 15. y 17. y
5 6 5
800 x
5 x 5 x
6 x Exercises 4.2
19. (A) y (B) y 21. (A) y 1. Yes 3. No 13. y 15. y 17. IV; 18, 02,
6 6 10
5. No 7. Yes 8 118, 02, 16, 42
4
9. IV 11. I 19. I; 10, 162,
10 x 10 x 14 x
16, 42, 118, 02
6 x 6 x
(B) y 23. Let a = no. of applicants; a 6 10 21. Unbounded 23. Unbounded 25. Bounded 27. Bounded
10
25. Let h = no. of hours of practice per day; 29. Bounded 31. Unbounded 33. Bounded
y y (1, 4)
h Ú 2.5 27. Let s = success rate; 10 x 4y 17
10 x 4
s 6 20% 29. Let f = annual fee; f 7 $45,000 (0, 4) (0, 7)
3 (6, 0) 3x 4y 24
(0, 6)
31. Let e = enrollment; e … 30 x y 7 0, 17
x 4
(0, 0) 5 (4, 3)
(8, 0)
33. 2x + 3y = - 6; 2x + 3y Ú - 6 35. y = 3; y 6 3 37. 4x - 5y = 0; 2x 3y 12
4x y 8
10 x (0, 0) 4x
4x - 5y Ú 0 39. Let x = enrollment in finite mathematics; (7, 0)
(2, 0)
let y = enrollment in calculus; x + y 6 300 41. Let x = revenue;
y = cost; x … y - 20,000 43. Let x = no. of grams of protein in rice; 35. Unbounded 37. Bounded 39. Empty
y y y
let y = number of grams of proteins in potato; x 7 2y 12 y 15
6 (0, 6)
45. 100y 47. y 49. y 3x 2y 12 5x 4y 48 2x y 1
80 600 (2, 3) 9 , 16 (4, 7) 10
5 5 x 2y 12
3x 8y 24 x 2y 13
(10, 1)
x y 11
500 x (0, 0) 8 x (8, 2) x x
300 x x y 5 (8, 0) 12 14
80 x
51. y 53. The solution set is the empty set and has no graph. 41. Bounded 43. Unbounded 45. Bounded
700 x y 10 y 2x y 3 y x 2y 6
55. Let x = no. of acres planted y
2x y 21 15
x y 1
4
200
with corn. 8 x 2y 18 x y 0
(8, 5) y 6
Let y = no. of acres planted (2, 8)
(7, 6) x y 5
(2, 2)
400 x with soybeans. y 1
90x + 70y … 11,000, x Ú 0, 2 (11, 1) 14 x (2, 1)
(1, 1) (4, 1)
14 x
y Ú 0 200 x (3/2, 0)
x
Answers A-13
47. Bounded 51. (A) Let x = no. of days to operate plant A

y
16x 13y 120 y = no. of days to operate plant B
3x 4y 25
10 Maximize C = 1000x + 900y
(2.17, 6.56) subject to 20x + 25y Ú 200
5 (1.24, 5.32) 60x + 50y Ú 500
–4x 3y 11
(6.2, 1.6) x Ú 0, y Ú 0
x
0 5 10 Plant A: 5 days; Plant B: 4 days; min cost $8,600
49. (A) 3x + 4y = 36 and 3x + 2y = 30 intersect at 18, 32; (B) Plant A: 10 days; Plant B: 0 days; min cost $6,000
3x + 4y = 36 and x = 0 intersect at 10, 92; (C) Plant A: 0 days; Plant B: 10 days; min cost $8,000
3x + 4y = 36 and y = 0 intersect at 112, 02; 53. Let x = no. of buses
3x + 2y = 30 and x = 0 intersect at 10, 152; y = no. of vans
3x + 2y = 30 and y = 0 intersect at 110, 02; Maximize C = 1,200x + 100y
x = 0 and y = 0 intersect at 10, 02 subject to 40x + 8y Ú 400
(B) 18, 32, 10, 92, 110, 02, 10, 02 3x + y … 36
51. 6x + 4y … 108 y x Ú 0, y Ú 0
x + y … 24
25 7 buses, 15 vans; min cost $9,900
(6, 18)
x Ú 0 55. Let x = amount invested in the CD
y Ú 0 y = amount invested in the mutual fund
Maximize P = 0.05x + 0.09y
20 x subject to x + y … 60,000
53. (A) All production schedules in the feasible region that are on the graph y Ú 10,000
of 50x + 60y = 1,100 will result in a profit of $1,100. (B) There are many x Ú 2y
possible choices. For example, producing 5 trick skis and 15 slalom skis will x, y Ú 0
produce a profit of $1,150. All the production schedules in the feasible region $40,000 in the CD and $20,000 in the mutual fund; max return is $3,800
that are on the graph of 50x + 60y = 1,150 will result in a profit of $1,150. 57. (A) Let x = no. of gallons produced by the old process
55. 20x + 10y Ú 460 y y = no. of gallons produced by the new process
30x + 30y Ú 960 60 Maximize P = 60x + 20y
5x + 10y Ú 220 subject to 20x + 5y … 16,000
x Ú 0 (14, 18) 40x + 20y … 30,000
(20, 12)
y Ú 0 x Ú 0, y Ú 0
60 x Max P = $450 when 750 gal are produced using the old process
57. 10x + 20y … 800 y exclusively.
40
20x + 10y … 640 (16, 32)
(B) Max P = $380 when 400 gal are produced using the old process and
x Ú 0 700 gal are produced using the new process.
y Ú 0 (C) Max P = $288 when 1,440 gal are produced using the new process
exclusively.
40 x 59. (A) Let x = no. of bags of brand A
Exercises 4.3 y = no. of bags of brand B
Maximize N = 8x + 3y
1. Max Q = 154; Min Q = 0 3. Max Q = 120; Min Q = - 60 5. Max
subject to 4x + 4y Ú 1,000
Q = 0; Min Q = - 32 7. Max Q = 40; Min Q = - 48 9. Max P = 16
2x + y … 400
at x = 7 and y = 9 11. Max P = 84 at x = 7 and y = 9, at x = 0 and
x Ú 0, y Ú 0
y = 12, and at every point on the line segment joining the preceding two
150 bags brand A, 100 bags brand B;
points. 13. Min C = 32 at x = 0 and y = 8 15. Min C = 36 at x = 4
Max nitrogen = 1,500 lb
and y = 3 17. Max P = 78 at x = 0 and y = 6 19. Min C = 35 at
(B) 0 bags brand A, 250 bags brand B;
x = 0 and y = 5 21. Max P = 200 at x = 4 and y = 12 23. Min z = 10
9 16 Min nitrogen = 750 lb
at x = 0 and y = 5 25. Min z = 218 at x = and y =
5 5 61. Let x = no. of cubic yards of mix A 63. Let x = no. of mice used
27. Max P = 180 at x = 6 and y = 2 29. Min P = 20 at x = 0 and y = no. of cubic yards of mix B y = no. of rats used
y = 2; Max P = 150 at x = 5 and y = 0 31. Feasible region empty; Minimize C = 30x + 35y Maximize P = x + y
no optimal solutions 33. Min P = 180 at x = 6 and y = 2 35. Max subject to 20x + 10y Ú 460 subject to 10x + 20y … 800
P = 20,000 at x = 0 and y = 1,000 37. Max P = 5,507 at x = 6.62 and 30x + 30y Ú 960 20x + 10y … 640
y = 4.25 39. Max z = 2 at x = 4 and y = 2; min z does not exist 41. 5x + 10y Ú 220 x Ú 0, y Ú 0
5 6 a 6 30 43. a 7 30 45. a = 30 47. a = 0 x Ú 0, y Ú 0 48; 16 mice, 32 rats
49. (A) Let: x = no. of trick skis (B) Max profit decreases 20 yd3 A, 12 yd3 B; $1,020
y = no. of slalom skis to $720 when 18 trick
Maximize P = 40x + 30y skis and no slalom skis Chapter 4 Review Exercises
subject to 6x + 4y … 108 are produced. y y
1. (4.1) 2. (4.1) 3. Bounded (4.2)
x + y … 24 (C) Max profit increases to 5 10 y
x Ú 0, y Ú 0 $1,080 when no trick 25
Max profit = $780 when 6 trick skis skis and 24 slalom skis 5 x 10 x
Solution
and 18 slalom skis are produced. are produced. (0, 10) region
(18, 0)
(0, 0) 25 x
A-14 Answers
4. Unbounded (4.2) 5. Bounded (4.2) 6. Unbounded (4.2) 39. 1x1, x2, s1, s2, s3 2 = 18, 16, 0, - 2, 02
y x2 x2
Solution
41. 4x1 + 5x2 + s1 = 20
120 5 (0, 3.3) 10 Solution
region (0, 9)
(1.4, 2.9)
region x1 x2 s1
(0, 75)
Solution 0 0 20 Feasible
region (2, 3)
(80, 0) (0, 0) (2.3, 0)
0 4 0 Feasible
(8, 0)
120 x 5 x1 10 x1 5 0 0 Feasible
7. 2x + 3y = 2, 2x + 3y Ú 2 (4.1) 8. 4x + y = 8; 4x + y Ú 8 (4.1) 43. x1 + x2 + s1 = 6
9. Max P = 18 at x = 6 and y = 0 (4.3) 10. Max P = 18.08 at x = 17.54 x1 + 4x2 + s2 = 12
and y = 0 (4.3) 11. Max P = 26 at x = 2 and y = 5 (4.3) 12. Min C = 51
x1 x2 s1 s2
at x = 3 and y = 9 (4.3) 13. Max P = 84 at x = 12 and y = 12 (4.3)
14. Let x = no. of calculator boards. (4.1)
0 0 6 12 Feasible
y = no. of toaster boards 0 6 0 -12 Not feasible
(A) 4x + 3y … 300, x Ú 0, y Ú 0 (B) 2x + y … 120, x Ú 0, y Ú 0 0 3 3 0 Feasible
y y 6 0 0 6 Feasible
120 160
12 0 -6 0 Not feasible
4 2 0 0 Feasible
45. 2x1 + 5x2 + s1 = 20
120 x 120 x x1 + 2x2 + s2 = 9
15. (A) Let x = no. of regular sails x1 x2 s1 s2
y = no. of competition sails 0 0 20 9 Feasible
Maximize P = 100x + 200y 0 4 0 1 Feasible
subject to 2x + 3y … 150 0 9>2 -5>2 0 Not feasible
4x + 10y … 380
10 0 0 -1 Not feasible
x, y Ú 0
9 0 2 0 Feasible
Max P = $8,500 when 45 regular and 20 competition sails are
5 2 0 0 Feasible
produced.
(B) Max profit increases to $9,880 when 38 competition and no regular 47. x1 + 2x2 + s1 = 24
sails are produced. x1 + x2 + s2 = 15
(C) Max profit decreases to $7,500 when no competition and 75 regular 2x1 + x2 + s3 = 24
sails are produced. (4.3) x1 x2 s1 s2 s3
16. (A) Let x = no. of grams of mix A 0 0 24 15 24 Feasible
y = no. of grams of mix B 0 12 0 3 12 Feasible
Minimize C = 0.04x + 0.09y
0 15 -6 0 9 Not feasible
0 24 -24 -9 0 Not feasible
2x + 4y Ú 800
24 0 0 -9 -24 Not feasible
4x + 5y Ú 1,150
x, y Ú 0 15 0 9 0 -6 Not feasible
Min C = $16.50 when 300 g mix A and 50 g mix B are used 12 0 12 3 0 Feasible
(B) The minimum cost decreases to $13.00 when 100 g mix A and 150 g 6 9 0 0 3 Feasible
mix B are used 8 8 0 -1 0 Not feasible
(C) The minimum cost increases to $17.00 when 425 g mix A and no mix 9 6 3 0 0 Feasible
B are used (4.3)
49. Corner points: 10, 02, 10, 42, 15, 02
x2
Chapter 5 4 (0, 4)
Exercises 5.1
(5, 0)
1. 56 3. 55 5. 21 7. 70 9. 1x1, x2, s1, s2 2 = 10, 2, 0, 22 (0, 0) x1
5
11. 1x1, x2, s1, s2 2 = 18, 0, - 6, 02
51. Corner points: 10, 02, 10, 32, 16, 02, 14, 22
13. 2x1 + 3x2 + s1 = 9 x2
6x1 + 7x2 + s2 = 13
15. 12x1 - 14x2 + s1 = 55 (0, 3)
3
19x1 + 5x2 + s2 = 40 (4, 2)
- 8x1 + 11x2 + s3 = 64 (6, 0)

17. 6x1 + 5x2 + s1 = 18 (0, 0) 6 x1
19. 4x1 - 3x2 + s1 = 12 53. Corner points: 10, 02, 10, 42, 19, 02, 15, 22
5x1 + 2x2 + s2 = 25 x2
- 3x1 + 7x2 + s3 = 32
4 (0, 4)
2x1 + x2 + s4 = 9
(5, 2)
21. s1, s2 23. x1, s2 25. (A), (B), (E), (F) 27. Max P = 40 at x1 = 0,
x2 = 8 29. The points below the line 2x1 + 3x2 = 24 31. x2, s1, s3 (9, 0)
33. x2, s3 35. (C), (D), (E), (F) 37. 1x1, x2, s1, s2, s3 2 = 112, 0, 12, 6, 02 (0, 0) 9 x1
Answers A-15
59. Max P = 50 at x1 = 5, x2 = 0 61. Max P = 100 at x1 = 4, x2 = 2 27. Max P = 190 at x1 = 10, x2 = 40, x3 = 0, s1 = 0, s2 = 0, s3 = 10
63. Max P = 225 at x1 = 9, x2 = 0 65. Max P = 540 at x1 = 6, x2 = 9 29. Max P = 26,000 at x1 = 400 and x2 = 600
67. Every point with coordinates 10, x2 2, where x2 Ú 0, satisfies the problem 31. Max P = 450 at x1 = 0, x2 = 180, and x2 = 30
constraint. So the feasible region is unbounded. 69. Every point with coordi- 33. Max P = 88 at x1 = 24 and x2 = 8 x2
15 (4, 14)
nates 1x1, x2 2, where x1 = x2 Ú 0, satisfies both problem constraints. So the (12, 12)
(0, 14)
feasible region is unbounded. 71. If 1x1, x2 2 is a point in the feasible region,
(24, 8)
then 0 … x2 … 3000 and 0 … x1 … 7000 - x2 6 7000 - 3000 … 4000,
so the feasible region is bounded; Max P = 62000 at x1 = 4000, x2 = 3000.
73. 10C4 = 210 75. 72C30 ≈ 1.64 * 1020 (0, 0) 40 x1
35. No solution 37. Choosing either col. produces the same optimal
Exercises 5.2 solution: max P = 13 at x1 = 13 and x2 = 10 39. Choosing col. 1: max
P = 60 at x1 = 12, x2 = 8, and x3 = 0. Choosing col. 2: max P = 60 at
1. (A) Basic: x2, s1, P; nonbasic: x1, s2 (B) x1 = 0, x2 = 12, s1 = 15,
x1 = 0, x2 = 20, and x3 = 0.
s2 = 0, P = 50 (C) Additional pivot required 3. (A) Basic: x2, x3,
41. Let x1 = no. of A components
s3, P; nonbasic: x1, s1, s2 (B) x1 = 0, x2 = 15, x3 = 5, s1 = 0, s2 = 0,
x2 = no. of B components
s3 = 12, P = 45 (C) No optimal solution
x3 = no. of C components
5. Enter
T Maximize P = 7x1 + 8x2 + 10x3
x1 x2 s1 s2 P subject to 2x1 + 3x2 + 2x3 … 1,000
Exit S s1 1 4 1 0 0 4 x1 + x2 + 2x3 … 800
s2 C 3 5 0 1 0 3 24 S x1, x2, x3 Ú 0
P -8 -5 0 0 1 0 200 A components, 0 B components, and 300 C components; max profit is $4,400
x1 1 4 1 0 0 4 43. Let x1 = amount invested in government bonds
x2 = amount invested in mutual funds
∼ s2 C 0 -7 -3 1 0 3 12 S
x3 = amount invested in money markey funds
P 0 27 8 0 1 32
Maximize P = 0.08x1 + 0.13x2 + 0.15x3
7. Enter subject to x1 + x2 + x3 … 100,000
T - x1 + x2 + x3 … 0
x1 x2 s1 s2 s3 P
x1, x2, x2 Ú 0
Exit S x2 2 1 1 0 0 0 4
$50,000 in government bonds, $0 in mutual funds, and $50,000 in money
s2 3 0 1 1 0 0 4 8
D T market funds; max return is $11,500
s3 0 0 2 0 1 0 2
45. Let x1 = no. of ads placed in daytime shows
P -4 0 -3 0 0 1 5
1 1
x2 = no. of ads placed in prime-time shows
x1 1 2 2 0 0 0 2 x3 = no. of ads placed in late-night shows
s2 0 - 32 - 12 1 0 0
∼ ≥ 4 2¥ Maximize P = 14,000x1 + 24,000x2 + 18,000x3
s3 0 0 2 0 1 0 2 subject to x1 + x2 + x3 … 15
P 0 2 -1 0 0 1 13 1,000x1 + 2,000x2 + 1,500x3 … 20,000
9. (A) 2x1 + x2 + s1 = 10 x1, x2, x3 Ú 0
x1 + 3x2 + s2 = 10 10 daytime ads, 5 prime-time ads, and 0 late-night ads; max no. of potential
- 15x1 - 10x2 + P = 0 customers is 260,000
x1, x2, s1, s2 Ú 0 47. Let x1 = no. of colonial houses
(B) Enter x2 = no. of split-level houses
T x3 = no. of ranch houses
x1 x2 s1 s2 P Maximize P = 20,000x1 + 18,000x2 + 24,000x3
Exit S s1 2 1 1 0 0 10 subject to 1
2 x1 +
1
2 x2 + x3 … 30
s2 C 1 3 0 1 0 3 10 S 60,000x1 + 60,000x2 + 80,000x3 … 3,200,000
P - 15 - 10 0 0 1 0 4,000x1 + 3,000x3 + 4,000x3 … 180,000
x1, x2, x3 Ú 0
(C) Max P = 80 at x1 = 4 and x2 = 2 20 colonial, 20 split-level, and 10 ranch houses; max profit is $1,000,000
11. (A) 2x1 + x2 + s1 = 10 49. The model is the same as the model for Problem 43 except that
x1 + 3x2 + s2 = 10 P = 17,000x1 + 18,000x2 + 24,000x3.
- 30x1 - x2 + P = 0 0 colonial, 40 split-level, and 10 ranch houses; max profit is $960,000; 20,000
x1, x2, s1, s2 Ú 0 labor-hours are not used
(B) Enter 51. The model is the same as the model for Problem 43 except that
T P = 25,000x1 + 18,000x2 + 24,000x3.
x1 x2 s1 s2 P
Exit S s1 45 colonial, 0 split-level, and 0 ranch houses; max profit is $1,125,000;
2 1 1 0 0 10
7.5 acres of land and $500,000 of capital are not used
s2 C 1 3 0 1 0 3 10 S
P - 30 - 1 0 0 1 0 53. Let x1 = no. of grams of food A Maximize P = 3x1 + 4x2 + 5x3
x2 = no. of grams of food B subject to x1 + 3x2 + 2x3 … 30
(C) Max P = 150 at x1 = 5 and x2 = 0 x3 = no. of grams of food C 2x1 + x2 + 2x3 … 24
13. Max P = 260 at x1 = 2 and x2 = 5 15. No optimal solution exists. x1, x2, x3 Ú 0
17. Max P = 7 at x1 = 3 and x2 = 5 19. Max P = 90 at x1 = 6 and 0 g food A, 3 g food B, and 10.5 g food C; max protein is 64.5 units
x2 = 0 21. No optimal solution exists. 23. Max P = 58 at x1 = 12, x2 = 0,
and x3 = 2 25. Max P = 17 at x1 = 4, x2 = 3 and x3 = 0
A-16 Answers
55. Let x1 = no. of undergraduate students 43. (A) Minimize C = 3x1 + x2 + 5x3
x2= no. of graduate students subject to - 2x1 + 6x2 + x3 Ú - 10
x3= no. of faculty members - 5x1 + x2 + 4x3 Ú 15
Maximize P = 18x1 + 25x2 + 30x3 x1, x2, x3 Ú 0
subject to x1 + x2 + x3 … 20 (B) Maximize P = - 10y1 + 15y2
100x1 + 150x2 + 200x3 … 3,200 subject to - 2y1 - 5y2 … 3
x1, x2, x3 Ú 0 6y1 + y2 … 1
0 undergraduate students, 16 graduate students, and 4 faculty members; y1 + 4y2 … 5
max no. of interviews is 520 y1, y2 Ú 0
(C) Yes
Exercises 5.3 45. Min C = 44 at x1 = 0, x2 = 3, and x3 = 5
2 5 47. Min C = 166 at x1 = 0, x2 = 12, x3 = 20, and x4 = 3
2 -5 6 49. Let x1 = no. of hours Oven A is operated
1 6 9 1 2
-5 2 3 1 9 x2 = no. of hours Oven B is operated
1. c d 3. £ 7 3 4§ 5. E - 6 0U 7. C S
3 7 0 8 0 x3 = no. of hours Oven C is operated
8 5 2 0 1
-1 4 -7 Minimize C = 8x1 + 6x2 + 20x3
-1 3
subject to 15x1 + 15x2 + 30x3 Ú 105
9. (A) Maximize P = 4y1 + 5y2
30x1 + 15x2 + 15x3 Ú 60
subject to y1 + 2y2 … 8
x1, x2, x3 Ú 0
3y1 + y2 … 9
Oven B 7 hr per day, Ovens A and C not used; min cost is $42
y1, y2 Ú 0
51. The model is the same as the model for Problem 49 except that the second
(B) y1 + 2y2 + x1 = 8
constraint (multigrain bread) is 30x1 + 15x2 + 15x3 Ú 150. Oven A 3 hr per
3y1 + y2 + x2 = 9
day, Oven B 4 hr per day, Oven C not used; min cost is $48.
- 4y1 - 5y2 + P = 0
53. The model is the same as the model for Problem 49 except that the second
(C) y1 y2 x1 x2 P
constraint (multigrain bread) is 30x1 + 15x2 + 15x3 Ú 180. Oven A 5 hr per
1 2 1 0 0 8
day, Oven B 2 hr per day, Oven C not used; min cost is $52.
C 3 1 0 1 0 3 9S
55. Let x1 = no. of ounces of food L
-4 -5 0 0 1 0
x2 = no. of ounces of food M
x3 = no. of ounces of food N
11. (A) Max P = 121 at y1 = 3 and y2 = 5 (B) Min C = 121 at x1 = 1
Minimize C = 20x1 + 24x2 + 18x3
and x2 = 2
subject to 20x1 + 10x2 + 10x3 Ú 300
13. (A) Maximize P = 13y1 + 12y2
10x1 + 10x2 + 10x3 Ú 200
10x1 + 15x2 + 10x3 Ú 240
y1 + y2 … 2
x1, x2, x3 Ú 0
y1, y2 Ú 0
10 oz L, 8 oz M, 2 oz N; min cholesterol intake is 428 units
(B) Min C = 26 at x1 = 0 and x2 = 13
57. Let x1 = no. of employees provided cab from East branch to South
15. (A) Maximize P = 15y1 + 8y2
x2 = no. of employees provided cab from East branch to North
subject to 2y1 + y2 … 7
x3 = no. of employees provided cab from West branch to South
3y1 + 2y2 … 12
x4 = no. of employees provided cab from West branch to North
y1, y2 Ú 0
Minimize C = 10x1 + 8x2 + 6x3 + 2x4
(B) Min C = 54 at x = 6 and x2 = 1
subject to x1 + x2 Ú 50
x3 + x4 Ú 30
subject to 2y1 - 2y2 … 11
x1 + x3 … 70
y1 + 3y2 … 4
x2 + x4 … 20
y1, y2 Ú 0
x1, x2, x3, x4 Ú 0
(B) Min C = 32 at x1 = 0
50 employees are provided cab from East branch to South, 10 from West
and x2 = 8
branch to South, 20 from West branch to North, zero from East branch to
North; min cost is $600.
subject to - 3y1 + y2 … 7
y1 - 2y2 … 9
y1, y2 Ú 0 Exercises 5.4
(B) No optimal solution exists. 1. (A) Maximize P = 5x1 + 2x2 - Ma1
21. Min C = 24 at x1 = 8 and x2 = 0 23. Min C = 20 at x1 = 0 and subject to x1 + 2x2 + s1 = 12
x2 = 4 25. Min C = 140 at x1 = 14 and x2 = 0 27. Min C = 44 at x1 + x2 - s2 + a1 = 4
x1 = 6 and x2 = 2 29. Min C = 100 at x1 = 0 and x2 = 4 31. Min x1, x2, s1, s2, a1 Ú 0
C = 24 at x1 = 0, x2 = 0, and x3 = 2 33. Min C = 52 at x1 = 0, x2 = 0 (B) x1 x2 s1 s2 a1 P
and x3 = 52 35. No optimal solution exists. 37. 2 variables and 4 problem 1 2 1 0 0 0 12
constraints 39. 2 constraints and any no. of variables C 1 1 0 -1 1 0 † 4S
41. (A) Maximize P = 7y1 + 10y2 -M - 5 -M - 2 0 M 0 1 - 4M
(C) x1 = 12, x2 = 0, s1 = 0, s2 = 8, a1 = 0, P = 60
2y1 + 6y2 … - 1
(D) Max P = 60 at x1 = 12 and x2 = 0
y1, y2 Ú 0
(B) No
Answers A-17
3. (A) Maximize P = 3x1 + 5x2 - Ma1 37. Let x1 = no. of cubic yards of mix A
subject to 2x1 + x2 + s1 = 8 x2 = no. of cubic yards of mix B
x1 + x2 + a1 = 6 x3 = no. of cubic yards of mix C
x1, x2, s1, a1 Ú 0 Maximize P = 12x1 + 16x2 + 8x3
(B) x1 x2 s1 a1 P subject to 12x1 + 8x2 + 16x3 … 700
2 1 1 0 0 8 16x1 + 8x2 + 16x3 Ú 800
£ 1 1 0 1 0 6 § x1, x2, x3 Ú 0
- M - 3 - M - 5 0 0 1 - 6M 25 yd3 A, 50 yd3 B, 0 yd3 C; max is 1,100 lb
(C) x1 = 0, x2 = 6, s1 = 2, a1 = 0, P = 30 39. Let x1 = no. of car frames produced at the Milwaukee plant
(D) Max P = 30 at x1 = 0 and x2 = 6 x2 = no. of truck frames produced at the Milwaukee plant
5. (A) Maximize P = 4x1 + 3x2 - Ma1 x3 = no. of car frames produced at the Racine plant
subject to - x1 + 2x2 + s1 = 2 x4 = no. of truck frames produced at the Racine plant
x1 + x2 - s2 + a1 = 4 Maximize P = 50x1 + 70x2 + 50x3 + 70x4
x1, x2, s1, s2, a1 Ú 0 subject to x1 + x3 … 250
(B) x1 x2 s1 s2 a1 P x2 + x4 … 350
-1 2 1 0 0 0 2 x1 + x2 … 300
C 1 1 0 -1 1 0 † 4S x3 + x4 … 200
-M - 4 -M - 3 0 M 0 1 - 4M 150x1 + 200x2 … 50,000
(C) No optimal solution exists. 135x3 + 180x4 … 35,000
(D) No optimal solution exists. x1, x2, x3, x4 Ú 0
7. (A) Maximize P = 5x1 + 10x2 - Ma1 41. Let x1 = no. of barrels of A used in regular gasoline
subject to x1 + x2 + s1 = 3 x2 = no. of barrels of A used in premium gasoline
2x1 + 3x2 - s2 + a1 = 12 x3 = no. of barrels of B used in regular gasoline
x1, x2, s1, s2, a1 Ú 0 x4 = no. of barrels of B used in premium gasoline
(B) x1 x2 s1 s2 a1 P x5 = no. of barrels of C used in regular gasoline
1 1 1 0 0 0 3 x6 = no. of barrels of C used in premium gasoline
Maximize P = 10x1 + 18x2 + 8x3 + 16x4 + 4x5 + 12x6
C 2 3 0 -1 1 0 † 12 S
subject to x1 + x2 … 40,000
- 2M - 5 - 3M - 10 0 M 0 1 - 12M
x3 + x4 … 25,000
(C) x1 = 0, x2 = 3, s1 = 0, s2 = 0, a1 = 3, P = - 3M + 30
x5 + x6 … 15,000
(D) No optimal solution exists.
x1 + x3 + x5 Ú 30,000
9. Min P = 1 at x1 = 3 and x2 = 5; max P = 16 at x1 = 8 and x2 = 0
x2 + x4 + x6 Ú 25,000
11. Max P = 44 at x1 = 2 and x2 = 8 13. No optimal solution exists.
- 5x1 + 5x3 + 15x5 Ú 0
15. Min C = - 9 at x1 = 0, x2 = 74, and x3 = 34 17. Max P = 32 at x1 = 0,
- 15x2 - 5x4 + 5x6 Ú 0
x2 = 4, and x3 = 2 19. Max P = 65 at x1 = 35 15
2 , x2 = 0, and x3 = 2
x1, x2, x3, x4, x5, x6 Ú 0
21. Max P = 120 at x1 = 20, x2 = 0, and x3 = 20 23. Problem 5:
43. Let x1 = percentage invested in high@tech funds
unbounded feasible region:
x2 = percentage invested in global funds
Problem 7: empty feasible region:
x2 x2 x3 = percentage invested in corporate bonds
8 8 x4 = percentage invested in municipal bonds
x5 = percentage invested in CDs
(2, 2) Maximize P = 0.11x1 + 0.1x2 + 0.09x3 + 0.08x4 + 0.05x5
(4, 0) subject to x1 + x2 + x3 + x4 + x5 = 1
8 x1 8 x1 2.7x1 + 1.8x2 + 1.2x3 + 0.5x4 … 1.8
25. Min C = - 30 at x1 = 0, x2 = 34 , and x3 = 0 27. Max P = 17 at x5 Ú 0.2
x1 = 495 , x2 = 0, and x3 = 5
22
29. Min C = 135 15 3
2 at x1 = 4 , x2 = 4 , and
x1, x2, x3, x4, x5 Ú 0
x3 = 0 31. Max P = 372 at x1 = 28, x2 = 4, and x3 = 0 45. Let x1 = no. of ounces of food L
33. Let x1 = no. of ads placed in the Sentinel x2 = no. of ounces of food M
x2 = no. of ads placed in the Journal x3 = no. of ounces of food N
x3 = no. of ads placed in the Tribune Minimize C = 0.4x1 + 0.6x2 + 0.8x3
Minimize C = 200x1 + 200x2 + 100x3 subject to 30x1 + 10x2 + 30x3 Ú 400
subject to x1 + x2 + x3 … 10 10x1 + 10x2 + 10x3 Ú 200
2,000x1 + 500x2 + 1,500x3 Ú 16,000 10x1 + 30x2 + 20x3 Ú 300
x1, x2, x3 Ú 0 8x1 + 4x2 + 6x3 … 150
2 ads in the Sentinel, 0 ads in the Journal, 8 ads in the Tribune; min cost is 60x1 + 40x2 + 50x3 … 900
$1,200 x1, x2, x3 Ú 0
35. Let x1 = no. of bottles of brand A 47. Let x1 = no. of students from town A enrolled in school I
x2 = no. of bottles of brand B x2 = no. of students from town A enrolled in school II
x3 = no. of bottles of brand C x3 = no. of students from town B enrolled in school I
Minimize C = 0.6x1 + 0.4x2 + 0.9x3 x4 = no. of students from town B enrolled in school II
subject to 10x1 + 10x2 + 20x3 Ú 100 x5 = no. of students from town C enrolled in school I
2x1 + 3x2 + 4x3 … 24 x6 = no. of students from town C enrolled in school II
x1, x2, x3 Ú 0
0 bottles of A, 4 bottles of B, 3 bottles of C; min cost is $4.30
A-18 Answers
Minimize C = 4x1 + 8x2 + 6x3 + 4x4 + 3x5 + 9x6 18. y1 y2 x1 x2 P

subject to x1 + x2 = 500 1 2 1 0 0 5
x3 + x4 = 1,200 C 3 1 0 1 0 2 S (5.3)
x5 + x6 = 1,800 - 15 - 20 0 0 1 0
x1 + x3 + x5 … 2,000 19. Max P = 40 at y1 = 0 and y2 = 2 (5.2) 20. Min C = 40 at x1 = 0
x2 + x4 + x6 … 2,000 and x2 = 20 (5.3) 21. Max P = 26 at x1 = 2 and x2 = 5 (5.2)
x1 + x3 + x5 Ú 1,400 22. Maximize P = 10y1 + 15y2 + 3y3
x2 + x4 + x6 Ú 1,400 subject to y1 + y2 … 3
x1 … 300 y1 + 2y2 + y3 … 8
x2 … 300 y1, y2, y3 Ú 0 (5.3)
x3 … 720 23. Min C = 51 at x1 = 9 and x2 = 3 (5.3) 24. No optimal solution exists.
x4 … 720 (5.2) 25. Max P = 23 at x1 = 4, x2 = 1, and x3 = 0 (5.2) 26. Max P = 84
x5 … 1,080 at x1 = 0, x2 = 12, and x3 = 0 (5.1) 27. Two pivot operations (5.2)
x6 … 1,080 28. (A) Modified problem:
x1, x2, x3, x4, x5, x6 Ú 0 Maximize P = x1 + 3x2 - Ma1
Chapter 5 Review Exercises subject to x1 + x2 - s1 + a1 = 6
1. x2, s2 (5.1) 2. x2, s1 (5.1) 3. x1 = 14, x2 = 0, s1 = 4, s2 = 0 (5.1) x1 + 2x2 + s2 = 8
4. x1 = 6, x2 = 4, s1 = 0, s2 = 0 (5.1) 5. (A), (B), (E), and (F) (5.1) x1, x2, s1, s2, a1 Ú 0
6. The points above the line 2x1 + 5x2 = 32 (5.1) 7. Max P = 700 at (B) Preliminary simplex tableau:
x1 = 14 and x2 = 0 (5.1) 8. 84 (5.1) x1 x2 s1 a1 s2 P
9. 2x1 + x2 + s1 = 8 1 1 -1 1 0 0 6
x1 + 2x2 + s2 = 10 (5.1) C 1 2 0 0 1 0 3 8S
10. 2 basic and 2 nonbasic variables (5.1) -1 -3 0 M 0 1 0
11.
x1 x2 s1 s2 Feasible? Initial simplex tableau:
x1 x2 s1 a1 s2 P
0 0 8 10 Yes
1 1 -1 1 0 0 6
0 8 0 -6 No
C 1 2 0 0 1 03 8S
0 5 3 0 Yes
-M -1 -M -3 M 0 0 1 - 6M
4 0 0 6 Yes
10 0 -12 0 No (C) x1 = 4, x2 = 2, s1 = 0, a1 = 0, s2 = 0, P = 10 (D) Since a1 = 0,
2 4 0 0 Yes (5.1) the optimal solution to the original problem is Max P = 10 at x1 = 4 and
12. Enter x2 = 2 (5.4)
T 29. (A) Modified problem:
x1 x2 s1 s2 P
Exit S s1 Maximize P = x1 + x2 - Ma1
2 1 1 0 0 8
subject to x1 + x2 - s1 + a1 = 5
s2 C 1 2 0 1 0 3 10 S (5.2)
P -6 -2 x1 + 2x2 + s2 = 4
0 0 1 0
x1, x2, s1, s2, a1 Ú 0
13. Max P = 24 at x1 = 4 and x2 = 0 (5.2) (B) Preliminary simplex tableau:
14. Basic variables: x2, s2, s3, P; nonbasic variables: x1, x3, s1 x1 x2 s1 a1 s2 P
Enter 1 1 -1 1 0 0 5
T C 1 2 0 0 1 0 3 4S
x1 x2 x3 s1 s2 s3 P -1 -1 0 M 0 1 0
x2 2 1 3 - 1 0 0 0 20 Initial simplex tableau:
s2 3 0 4 1 1 0 0 4 30
D T x1 x2 s1 a1 s2 P
Exit S s3 2 0 5 2 0 1 0 10
1 1 -1 1 0 0 5
P -8 0 -5 3 0 0 1 50
C 1 2 0 0 1 0 3 4S
x1 x2 x3 s1 s2 s3 P -M - 1 -M - 1 M 0 0 1 - 5M
x2 0 1 -2 -3 0 -1 0 10
∼ s2 D 0 0 - 72 -2 1 - 32 0 4 15 T (5.2) (C) x1 = 4, x2 = 0, s1 = 0, s2 = 0, a1 = 1, P = - M + 4
5 1
x1 1 0 2 1 0 2 0 5 (D) Since a1 ≠ 0, the original problem has no optimal solution. (5.4)
P 0 0 15 11 0 4 1 90
15. (A) x1 = 0, x2 = 2, s1 = 0, s2 = 5, P = 12; additional pivoting required 30. Maximize P = 2x1 + 3x2 + x3 - Ma1 - Ma2
(B) x1 = 0, x2 = 0, s1 = 0, s2 = 7, P = 22; no optimal solution exists subject to x1 - 3x2 + x3 + s1 = 7
(C) x1 = 6, x2 = 0, s1 = 15, s2 = 0, P = 10; optimal solution (5.2) x1 + x2 - 2x3 - s2 + a1 = 2
16. Maximize P = 15y1 + 20y2 3x1 + 2x2 - x3 + a2 = 4
subject to y1 + 2y2 … 5 x1, x2, s1, s2, a1, a2 Ú 0 (5.4)
3y1 + y2 … 2 31. The basic simplex method with slack variables solves standard maximiza-
y1, y2 Ú 0 (5.3) tion problems involving … constraints with nonnegative constants on the
17. y1 + 2y2 + x1 = 5 right side. (5.2) 32. The dual problem method solves minimization problems
3y1 + y2 + x2 = 2 with positive coefficients in the objective function. (5.3) 33. The big M
- 15y1 - 20y2 + P = 0 (5.3) method solves any linear programming problem. (5.4)
Answers A-19
34. Max P = 36 at x1 = 6, x2 = 8 (5.2)

Chapter 6
x2
(0, 10)
(2, 10) Exercises 6.1
(6, 8)
8 1. 1, 2, 4, 5, 10, 20 3. 11, 22, 33, 44, 55 5. 23, 29 9. 91 is not prime; true
11. 91 is prime and 91 is odd; false 13. If 91 is odd, then 91 is prime; false
15. If the moon is a cube, then rain is wet; true 17. The moon is a cube or
(0, 0) 10 x1 rain is wet; true 19. If rain is not wet, then the moon is not a cube; true
35. Min C = 15 at x1 = 3 and x2 = 3 (5.3) 36. Min C = 15 at x1 = 3 21. Disjunction; true 23. Conditional; false 25. Negation; false
and x2 = 3 (5.4) 37. Min C = 9,960 at x1 = 0, x2 = 240, x3 = 400, and 27. Disjunction; true 29. Converse: If triangle ABC is equiangular, then
x4 = 60 (5.3) triangle ABC is equilateral. Contrapositive: If triangle ABC is not equiangular,
38. (A) Let x1 = amount invested in oil stock then triangle ABC is not equilateral. 31. Converse: If f1x2 is a decreasing
x2 = amount invested in steel stock function, then f1x2 is a linear function with negative slope. Contrapositive:
x3 = amount invested in government bonds If f1x2 is not a decreasing function, then f1x2 is not a linear function with
Maximize P = 0.12x1 + 0.09x2 + 0.05x3 negative slope. 33. Converse: If n is an integer that is a multiple of 2 and a
subject to x1 + x2 + x3 … 150,000 multiple of 4, then n is an integer that is a multiple of 8. Contrapositive: If n is
x1 … 50,000 an integer that is not a multiple of 2 or not a multiple of 4, then n is an integer
x1 + x2 - x3 … 25,000 that is not a multiple of 8.
x1, x2, x3 Ú 0 35. p q ¬p ¿ q Contingency
Max return is $12,500 when $50,000 is invested in oil stock, $37,500 T T F
is invested in steel stock, and $62,500 in government bonds. (B) Max T F F
return is $13,625 when $87,500 is invested in steel stock and $62,500 in
F T T
government bonds. (5.2)
F F F
39. (A) Let x1 = no. of regular chairs
x2 = no. of rocking chairs
37. p q ¬p S q Contingency
x3 = no. of chaise lounges
T T T
Maximize P = 17x1 + 24x2 + 31x3
subject to x1 + 2x2 + 3x3 … 2,500 T F T
2x1 + 2x2 + 4x3 … 3,000 F T T
3x1 + 3x2 + 2x3 … 3,500 F F F
x1, x2, x3 Ú 0
Max P = $30,000 when 250 regular chairs, 750 rocking chairs, and 250 39. p q q ¿ 1p ¡ q2 Contingency
chaise lounges are produced. (B) Maximum profit increases to $32,750 T T T
when 1,000 regular chairs, 0 rocking chairs, and 250 chaise lounges are T F F
produced. (C) Maximum profit decreases to $28,750 when 125 regular F T T
chairs, 625 rocking chairs, and 375 chaise lounges are produced. (5.2) F F F
40. Let x1 = no. of motors shipped from factory A to plant X
x2 = no. of motors shipped from factory A to plant Y 41. p q p ¡ 1p S q2 Tautology
x3 = no. of motors shipped from factory B to plant X
T T T
x4 = no. of motors shipped from factory B to plant Y
T F T
Minimize C = 5x1 + 8x2 + 9x3 + 7x4
F T T
subject to x1 + x2 … 1,500
+ x3 + x4 … 1,000 F F T
x1 + x3 Ú 900
x2 + x4 Ú 1,200 43. p q p S 1p ¿ q2 Contingency
x1, x2, x3, x4 Ú 0 T T T
Min C = $13,100 when 900 motors are shiped from factory A to plant X, T F F
200 motors are shipped from factory A to plant Y, and 1,000 motors are F T T
shipped from factory B to plant Y (5.3) F F T
41. Let x1 = no. of lbs of long@grain rice used in brand A
x2 = no. of lbs of long@grain rice used in brand B 45. p q 1p S q2 S ¬ p Contingency
x3 = no. of lbs of wild rice used in brand A
T T F
x4 = no. of lbs of wild rice used in brand B
Maximize P = 0.8x1 + 0.5x2 - 1.9x3 - 2.2x4 T F T
subject to 0.1x1 - 0.9x3 … 0 F T T
0.05x2 - 0.95x4 … 0 F F T
x1 + x2 … 8,000
x3 + x4 … 500 47. p q ¬ p S 1p ¡ q2 Contingency
x1, x2, x3, x4 Ú 0 T T T
Max profit is $3,350 when 1,350 lb long-grain rice and 150 lb wild rice are T F T
used to produce 1,500 lb of brand A, and 6,650 lb long-grain rice and 350 lb F T T
wild rice are used to produce 7,000 lb of brand B. (5.2) F F F
A-20 Answers
27. 51, 3, 5, 7, 96 29. A′=51,56 31. 100 33. 60 35. 81 37. 37

49. p q 1 p S q2 ¡ 1 q S p2 Tautology 39. 18 41. 19 43. 82 45. (A) 51, 2, 3, 4, 66 (B) 51, 2, 3, 4, 66
T T T 47. 53, 56 49. Infinite 51. Finite
T F T 53. 55.
A B A B
F T T
F F T
51. p q 1¬ p ¿ q2 ¿ 1q S p2 Contradiction
T T F
T F F
F T F C C
F F F
57. 59. Disjoint 61. Not disjoint
A B
53. p q p ¡ q 63. True 65. False 67. False
T T T 69. False 71. False 73. (A) 2
T F T (B) 4 (C) 8 (D) 16 75. 85
F T T 77. 33 79. 14 81. 66 83. 54
F F F 85. 106 87. 0 89. 6
91. A+ , AB+
55. p q p ¿ q p ¡ q 93. A- , A+ , B+ , AB- , AB+ , O+
C 95. O+ , O- 97. B- , B+
T T T T
T F F T
F T F T
Exercises 6.3
F F F F
1. 13 3. 13 5. 9 7. (A) 4 ways: (B) 2 # 2 = 4 ways
COIN 1 COIN 2 COMBINED
57. p q ¬ p S 1q ¿ ¬ q2 OUTCOME OUTCOME OUTCOMES
H (H, H)
T T T H
T (H, T)
T F T
Start
F T F H (T, H)
T
F F F T (T, T)
9. (A) 12 combined outcomes:

59. p q ¬ p S 1p ¡ q2 p ¡ q 1 (B) 2 # 6 = 12 combined outcomes
2
T T T T H
3
4
T F T T 5
F T T T Start
6
1
F F F F 2
3
T
4
61. p q q ¿ 1p ¡ q2 q ¡ 1p ¿ q2 5
6
T T T T
T F F F 11. (A) 19 (B) 84 13. 60; 125; 80 15. (A) 8 (B) 144
17. n1A ¨ B′2 = 50, n1A ¨ B2 = 50, n1A′ ¨ B2 = 40, n1A′ ¨ B′2 = 60
F T T T
19. n1A ¨ B′2 = 5, n1A ¨ B2 = 30, n1A′ ¨ B2 = 55, n1A′ ¨ B′2 = 10
F F F F
21. n1A ¨ B′2 = 160, n1A ¨ B2 = 30, n1A′ ¨ B2 = 50, n1A′ ¨ B′2 = 60
23. n1A ¨ B′2 = 30, n1A ¨ B2 = 30, n1A′ ¨ B2 = 10, n1A′ ¨ B′2 = 10
63. p q p ¡ 1p S q2 p S 1p ¡ q2 25.
T T T T A A′ Totals
T F T T B 30 60 90
F T T T B′ 40 70 110
F F T T Totals 70 130 200
27.
65. p S ¬q K ¬p ¡ ¬q By (4) A A′ Totals
K ¬1p ¿ q2 By (6) B 30 45 75
67. ¬1p S q2 K ¬1¬p ¡ q2 By (4) B′ 5 20 25
K ¬1¬p2 ¿ ¬q By (6)
Totals 35 65 100
K p ¿ ¬q By (1)
69. p ¿ ¬q 71. Contingency 73. No 29.
A A′ Totals
B 39 17 56
Exercises 6.2
B′ 56 8 64
1. No 3. No 5. No 7. T 9. T 11. T 13. F 15. 52, 36
Totals 95 25 120
17. 51, 2, 3, 46 19. 51, 4, 7, 10, 136 21. ∅ 23. 55,- 56 25. 5 - 36
Answers A-21
31. 19. ∅ (6.2) 20. (A) 12 combined outcomes: (B) 6 # 2 = 12 (6.3)

A A′ Totals 1 H
T
B 0 145 145 2 H
T
B′ 110 45 155 H
3
Totals 110 190 300 T
H
33. (A) True (B) False 35. 5 # 3 # 4 # 2 = 120 37. 526 = 19,770,609,664
4
T
H
39. 10 # 9 # 8 # 7 # 6 = 30,240; 10 # 10 # 10 # 10 # 10 = 100,000; 10 # 9 # 9 # 9 # 9
5 T
= 65,610 41. 26 # 26 # 26 # 10 # 10 # 10 = 17,576,000; 26 # 25 # 24 # 10 # 9 # 8

H
6 T
= 11,232,000 43. No, the same 8 combined choices are available either 21. (A) 65 (B) 75 (C) 150 (D) 85 (6.3) 22. (A) 35 (B) 105 (C) 115
way. 45. z 47. x and z 49. 14 51. 14 (D) 45 (6.3) 23. 24 (6.4) 24. 360,360 (6.4) 25. 3,003 (6.4)
53. (A) 6 combined outcomes: (B) 3 # 2 = 6 26. 56 (6.4) 27. 6720 (6.4) 28. 9,295 (6.4)
29. 6 # 5 # 4 # 3 # 2 # 1 = 720 (6.3) 30. 6P6 = 6! = 720 (6.4)
A
H
P
A
Start M Contingency (6.1)
P 31. p q 1p S q2 ¿ 1q S p2
A
L T T T
P
55. 12 57. (A) 809 (B) 691 (C) 437 59. 1,570 61. (A) 102,000 T F F
(B) 689,000 (C) 1,470,000 (D) 1,372,000 63. (A) 12 classifications: F T F
U (B) 2 # 2 # 3 = 12 65. 2,905 F F T
S A
M O
U 32. p q p ¡ 1q S p2 Contingency (6.1)
N A
Start
O T T T
U
S A
T F T
F
O F T F
U
N A F F T
O
33. p q 1p ¡ ¬p2 S 1q ¿ ¬q2 Contradiction (6.1)

Exercises 6.4 T T F
1. 220 3. 252 5. 9,900 7. 40,320 9. 5,040 11. 720 13. 990 T F F
15. 70 17. 249,500 19. 1,287 21. 13,366,080 23. 0.1114 25. 0.2215 F T F
27. n1n - 12 29. n1n + 12 >2 31. Permutation 33. Combination F F F
35. Neither 37. 10P3 = 10 # 9 # 8 = 720 39. 7C3 = 35; 7P3 = 210
41. 26C6 = 230,230 43. 12C5 = 792 45. 48C3 = 17,296 47. 134 = 28,561 p q ¬q ¿ 1p S q2 Contingency (6.1)
34.
49. 3,744 51. 8C3 # 10C4 # 7C2 = 246,960 53. The numbers are the same
T T F
read up or down, since nCr = nCn-r. 55. True 57. False 59. True
T F F
61. (A) 8C2 = 28 (B) 8C3 = 56 (C) 8C4 = 70 63. 6P4 = 360
65. (A) 8P5 = 6,720 (B) 8C5 = 56 (C) 2 # 6C4 = 30 67. n1K ¨ H′2
F T F
= 120, n1K ¨ H2 = 72, n1K′ ¨ H2 = 435, n1K′ ¨ H′2 = 699 F F T
69. 24C12 = 2,704,156 71. (A) 24C3 = 2,024 (B) 19C3 = 969
73. (A) 40C8 = 76,904,685 (B) 14C3 # 15C4 # 11C1 = 5,465,460 35. p q ¬p S 1p S q2 Tautology (6.1)
75. (A) 6C3 # 5C2 = 200 (B) 6C4 # 5C1 = 75 (C) 6C5 = 6 T T T
(D) 11C5 = 462 (E) 6C4 # 5C1 + 6C5 = 81 77. 336; 512 79. 4P2 = 12 T F T
F T T
Chapter 6 Review Exercises F F T
1. 32 is not less than 24; true (6.1) 2. 23 is less than 32 or 34 is less than 43;
true (6.1) 3. 23 is less than 32 and 34 is less than 43; false (6.1) 4. If 23 is 36. p q ¬1p ¡ ¬q2 Contingency (6.1)
less than 32, then 34 is less than 43; false (6.1) 5. If 32 is less than 24, then 43 T T F
is less than 112; false (6.1) 6. If 43 is greater than 112, then 32 is less than 24; T F F
true (6.1) 7. T (6.2) 8. T (6.2) 9. F (6.2) 10. T (6.2)
F T T
11. Conditional; false (6.1) 12. Disjunction; true (6.1) 13. Conjunction;
F F F
false (6.1) 14. Negation; true (6.1) 15. Converse: If the square matrix A
does not have an inverse, then the square matrix A has a row of zeros. Contra- 37. Infinite (6.2) 38. Finite (6.2) 39. Infinite (6.2) 40. Infinite (6.2)
positive: If the square matrix A has an inverse, then the square matrix A does 41. Disjoint (6.2) 42. Not disjoint (6.2)
not have a row of zeros. (6.1) 16. Converse: If the square matrix A has an 43. (6.2)
inverse, then the square matrix A is an identity matrix. Contrapositive: If the A B
square matrix A does not have an inverse, then the square matrix A is not an
identity matrix. (6.1) 17. 51, 2, 3, 4, 56 (6.2) 18. 52, 3, 46 (6.2)
C
A-22 Answers
44. 5 children, 15 grandchildren, and 30 great grandchildren, for a total of 69. (A) Yes (B) Yes, because we would expect, on average, 20 heads in
50 descendants (6.3, 6.4) 45. 336; 512; 392 (6.3) 46. (A) 6P3 = 120 40 flips; P1H2 = 37 3 1 3
40 = .925; P1T2 = 40 = .075 71. 16 73. 16 75. 16
3
(B) 5C2 = 10 (6.4) 47. 25C12 = 25C13 = 5,200,300 (6.4) 48. N1 # N2 # N3 (6.3) 1 #
77. 2 79. 26C5 > 52C5 ≈ .0253 81. 12C3 40C3 > 52C6 ≈ 0.1068
49. 5 - 1, 0, 16 (6.2) 50. 51, 2, 3, 46 (6.2, 6.4) 83. 48C4 > 52C4 ≈ .7187 85. 13C2 # 13C2 # 26C3 > 52C7 ≈ .1182
8
51. 51, 4, 9, 16, 25, 36, 496 (6.2) 52. (A) 10P3 = 720 (B) 6P3 = 120 87. (A) 50 = .16 (B) 16 ≈ .167 (C) Answer depends on results of
(C) 10C3 = 120 (6.3, 6.4) 53. 33 (6.3) 54. x (6.3) 55. x, z, and w (6.3) simulation. 89. (A) Represent the outcomes H and T by 1 and 2,
56. True (6.4) 57. False (6.4) 58. True (6.4) respectively, and select 500 random integers from the integers 1 and 2.
(B) Answer depends on results of simulation. (C) Each is 12 = .5
59. p q p ¿ q (6.1) 91. (A) 12P4 ≈ .000 084 (B) 1>124 ≈ .000 048 93. (A) 6C3 # 5C2 > 11C5 ≈ .433
T T T (B) 6C4 # 5C1 > 11C5 ≈ .162 (C) 6C5 > 11C5 ≈ .013 (D) 1 6C4 # 5C1 + 6C52>11C5
T F F ≈ .175 95. (A) 1> 8P3 ≈ .0030 (B) 1>83 ≈ .0020
F T F 97. (A) 6C3 > 11C3 ≈ .1212 (B) 1 6C3 + 6C2 # 5C1 2 > 11C3 ≈ .5758
F F F
Exercises 7.2
60. p q pSq (6.1) 1. 13 3. 24 7
5. 27 7. .38 9. .77 11. .27 13. 14 15. 10 3 11
13 17. 52 19. 26
5 43 5 11
T T T 21. 26 23. 52 25. .76 27. .52 29. .6 31. .36 33. .49 35. 18 37. 36
9
T F F 39. (A) 35 ; 53 (B) 13 ; 31 (C) 23 ; 32 (D) 11 3 11 4
9 ; 11 41. (A) 11 (B) 18 (C) 5 or .8
F T T (D) .49 43. False 45. False 47. True 49. 1:1 51. 7:1 53. 1:2
F F T 55. 1:2 57. (A) 18 (B) $8 59. (A) .31; 31 3 11
69 (B) .6; 2 61. 26 63. 13
7
16 12 4
65. .78 67. 100 + 100 - 100 = .24 69. Either events A, B, and C are
mutually exclusive, or events A and B are not mutually exclusive and the
61. p q ¬p S 1q ¿ ¬q2 (6.1)
other pairs of events are mutually exclusive. 71. There are fewer calculator
T T T
steps, and, in addition, 365! produces an overflow error on many calculators,
T F T
while 365Pn does not produce an overflow error for many values of n.
F T F 73. P1E2 = 1 - 112 -12!n2!12n 77. (A) 11 7 18
50 + 50 = 50 = .36
F F F 6 5 11
(B) 36 + 36 = 36 ≈ .306 (C) Answer depends on results of simulation.
(B) P1C′ xL′2 = .15 81. (A) P1M1 hA2 = P1M1 2 + P1A2
79. (A) P1ChL2 = P1C2 + P1L2 - P1CxL2 = .45 + .75 - .35 = .85
p q p ¡ q ¬p S q
- P1M1 xA2 = .2 + .3 - .05 = .45 (B) P3 1M2 xA′2 h 1M3 xA′24
62. (6.1)
= P1M2 xA′2 + P1M3 xA′2 = .2 + .35 = .55 83. .83

T T T T
= .05 87. (A) P1L hN2 = .67; 33

T F T T
(B) P3 1CxB2 h 1OxB24 = .05; 19

50
F T T T 85. P1A″S2 = 1,000 67
1
F F F F
63. p q p ¿ 1p S q2 (6.1) Exercises 7.3

T T T 1. One answer is: 3. One answer is:
2 2
T F F 4 18 3
2 9
100 180 3
F T F 5 2
25 10
F F F 5 5
64. p q pSq 5. One answer is: 7. 1/13 9. 1/4 11. 1 13. 1/2
¬1p ¿ ¬q2 (6.1) 5
15. 2/9 17. 2/5 19. 3/10 21. 3/25
T T T T 315
7 23. .10 25. .03 27. .3 29. .1
T F F F 63 3
31. .2 33. 0 35. Dependent
9
F T T T 3
37. Independent 39. Dependent
F F T T
41. Dependent 43. (A) 12 (B) 21 12 2 8 ≈ .00781 45. (A) 14 (B) Dependent
65. 25 = 32; 6 (6.3) 66. Yes, it is both if r = 0 or r = 1. (6.4) 67. 120 47. (A) .24 (B) .34 49. (A) Independent and not mutually exclusive
(6.3) 68. (A) 610 (B) 390 (C) 270 (6.3) 69. 40C6 = 3,838,380 (6.4) (B) Dependent and mutually exclusive. 51. 1 12 21 12 2 = 14 ; 12 + 12 - 14 = 34
70. (A) 67! ≈ 3.647 * 1094 (6.4) 53. (A) 1 14 21 13 1 1 3
51 2 ≈ .0637 (B) 1 4 21 4 2 = .0625 55. (A) 13
(B) Independent 57. (A) Dependent (B) Independent
Chapter 7 59. 2
7 R2 P(R1 R2)
4
49
2 R1
Exercises 7.1 7
5 10
1 4 1 1 1 W2 P (R1 W2)
1. E 3. F 5. F 7. 3 9. 11. 0 13. 1 15.
15 17. 4 19.2 52 (A) Start 7 49
7
21. 13 23. 0 25. 12 27. (A) Reject; no probability can be negative 2
R2 P (W1 R2)
10
5 7 49
(B) Reject; P1J2 + P1G2 + P1P2 + P1S2 ≠ 1 (C) Acceptable
7 W1
29. P1G2 + P1J2 = .4 31. 18 33. 1>104 = .0001 35. 26C5 > 52C5 ≈ .025 5 25
W2 P(W1 W2)
37. 12C5 > 52C5 ≈ .000 305 39. S = 5All days in a year, 365, excluding 7 49
1
leap year}; 365 , assuming each day is as likely as any other day for a person
5
to be born 41. 1> 5P5 = 1>5! = .008 33 43. 36 45. 16 47. 79 49. 0
1 2 2 1 1 3
51. 3 53. 9 55. 3 57. 4 59. 4 61. 4 63. No 65. Yes 67. Yes
Answers A-23
1 1 41. (A)
6 R2 P(R1 R2)
21
2 R1 xi 0 1 2
7 5 pi 7 7 1
5 P (R1 W2)
W2 21 15 15 15
(B) Start 6 (B) .60
2 5
6
R2 P (W1 R2)
21 43. (A)
5
7 W1 xi –$5 $195 $395 $595
4 10
W2 P(W1 W2)
21 pi .985 .0149 .000 059 9 .000 000 06
6
(B) E1X2 ≈ - $2
61. (A) 45 20
49 (B) 21 63. False 65. True 67. False 69. True 71. 5
- $56
18 45. (A) - $56 (B) The value per game is 200 = - $0.28, compared with an
73. (A) .167 (B) .25 (C) .25
expected value of - $0.0526. (C) The simulated gain or loss depends on the
81. (A) H S Totals (B) P1Y2 = .6; results of the simulation; the expected loss is $26.32. 47. $36.27
Y .400 .200 .600 P1Y H2 = .400
.550 ≈ .727 49. Payoff table:
(C) P1S2 = .450; xi $9875 –$125
N .150 .250 .400
P1S Y2 = .200
.600 ≈ .333 pi .005 .995
Totals .550 .450 1.000
83. (A) .167 E1X2 = - $75
(B) .25 (C) .25 51. Site A, with E1X2 = $3.6 million 53. 1.54 55. - £1.34
85. (A) H H′ Totals (B) Dependent
R .09 .41 .50 (C) P1H R2 = .18 Chapter 7 Review Exercises
and P1H2 = .12; since
R′ .03 .47 .50 1. 13C5 > 52C5 ≈ .0005 (7.1) 2. 1> 10P2 ≈ .0111 (7.1)
P1H R2 7 P1C2, the red
Totals .12 .88 1.00 3. 1> 10P3 ≈ .0014; 1> 10C3 ≈ .0083 (7.1) 4. .05 (7.1)
dye should be banned.
5. Payoff table:
87. (A) P1A F2 = .130
.520 = .250; P1A F′2 =
.120
.480 = .250
xi –$2 –$1 $0 $1 $2
(B) P1A2 = .250 (C) No; no 89. .234
pi 1 1 1 1 1
5 5 5 5 5
Exercises 7.4 E1X2 = 0; game is fair (7.5)

1. 2/5 3. 3/2 5. 9/10 7. 1.621.72 = .42 9. 1.621.72+1.421.22 = .50 6. (A) .55 (B) .3 (7.2) 7. P1RhG2 = .8; odds for RhG are 8 to 2 (7.2)
5
11. .84 13. .417 15. .375 17. .28 19. .12 21. .60 23. 13 25. 13 8. 11 ≈ .455 (7.2) 9. .27 (7.3) 10. .20 (7.3) 11. .02 (7.3) 12. .03 (7.3)
(7.3) 17. Yes, since P1S xX2 = P1S2P1X2 (7.3) 18. .4 (7.3) 19. .2 (7.3)
1
13. .15 (7.3) 14. .1304 (7.3) 15. .1 (7.3) 16. No, since P1T Z2 ≠ P1T2
27. 5 29. 1 31. .25 33. .333 35. .50 37. .745
10 A
41. (A) True (B) True 43. .235 20. .3 (7.3) 21. .08 (7.3) 22. .18 (7.3) 23. .26 (7.3) 24. .31 (7.4)
1 B 101 50
2
9
45. 204 ≈ .4951 47. 101 ≈ .4950 25. .43 (7.4) 26. (A) 165
(B) 14 (C) As the sample in part (A) increases in
A 51
Start
10 49. 101 ≈ .5050 53. .913; .226 size, approximate empirical probabilities should approach the theoretical prob-
2 55. .091; .545; .364 57. .667; .000 412 abilities. (7.1) 27. No (7.1) 28. Yes (7.1)
A
1 5 59. .231; .036 61. .941; .0588
2 B 29. Payoff table:
3 A xi $5 - $4 $2
5
pi .25 .5 .25
Exercises 7.5
E1X2 = - 25¢; game is not fair (7.5)
1. 71 3. 75 5. 142 7. E1X2 = - 0.1 9. $43.75 11. $0.148 13. $25
30. (A) 49 (B) 13 (7.3) 31. (A) 13 2 4
; 2 to 11; (B) 13 ; 4 to 9;
15. Probability distribution: 17. Payoff table: 12
(C) 13 ; 12 to 1 (7.2) 32. (A) 1 to 8 (B) $8 (7.2)
xi 0 1 2 xi $1 –$1 33. (A) P12 heads2 = .21, P11 head2 = .48, P10 heads2 = .31
pi 1 1 1 pi 1 1
(B) P12 heads2 = .25, P11 head2 = .50, P10 heads2 = .25 (C) 2 heads,
4 2 4 2 2
E1X2 = 1 E1X2 = 0; game is fair 250; 1 head, 500; 0 heads, 250 (7.1, 7.5)
34. 16 ; since the dice are independent of each other, the 5th roll is independent
19. Payoff table:
of the preceding 4 rolls. (7.3)
xi –$3 –$2 –$1 $0 $1 $2
35. (A)
pi 1 1 1 1 1 1
6 6 6 6 6 6 xi 2 3 4 5 6 7 8 9 10 11 12
E1X2 = - 50¢; game is not fair pi 1 2 3 4 5 6 5 4 3 2 1
36 36 36 36 36 36 36 36 36 36 36
21. - $2.50 23. - $0.035; $0.035 25. $40. Let x = amount you should
(B) E1X2 = 7 (7.5)
lose if a 6 turns up. Set up a payoff table; then set the expected value of the
36. A = 511, 32, 12, 22, 13, 12, 12, 62, 13, 52, 14, 42, 15, 32, 16, 22, 16, 626;
game equal to zero and solve for x. 27. Win $1 29. - $0.154 31. $2.75
B = 511, 52, 12, 42, 13, 32, 14, 22, 15, 12, 16, 626;
33. A2; $210
1 1 1 7
35. Payoff table: P(A) = ; P1B2 = ; P1A¨B2 = ; P1A∪B2 = (7.2)
4 6 36 18
xi $35 –$1
37. (1) Probability of an event cannot be negative; (2) sum of probabilities
pi 1 37
38 38 of simple events must be 1; (3) probability of an event cannot be greater
E1X2 = -5.26¢ than 1. (7.1)
37. .002 38.
39. Payoff table:
A A′ Totals
xi $499 $99 $19 $4 –$1
B 15 30 45
pi .0002 .0006 .001 .004 .9942
B′ 35 20 55
E1X2 = -80¢
Totals 50 50 100 (7.2)
A-24 Answers
1 6
39. (A) .6 (B) .5 (7.3) 40. (A) 13 (B) Independent (7.3) 41. (A) 25 A B C
3 3 B .2
(B) 10 (7.3) 42. Part (B) (7.3) 43. (A) 1.2 (B) 1.2 (7.5) 44. (A) 5 .4
A .1 .4 .5
9 .5
(B) 13 (C) 58 (D) 14 (7.3, 7.4) 45. No (7.3) 46. (A) 13C5 > 52C5 47. No 49. .1 .3 .2 B C .5 .2 .3 S
(B) 13C3 # 13C2 > 52C5 (7.1) 47. 2C2 # 18C2 > 20C4 ≈ .0316 (7.1) 48. Events
A
S and H are mutually exclusive. Hence, P1S xH2 = 0, while P1S2 ≠ 0 and
.7 C .7 .2 .1
P1H2 ≠ 0. Therefore, P1S xH2 ≠ P1S2P1H2, which implies S and H

.5 C .1
8 9
are dependent. (7.3) 49. (A) 50 = .16 (B) 36 = .25 (C) The empirical
51. a = .5, b = .6, c = .7 53. a = .7, b = 1, c = .2 55. No
probability depends on the results of the simulation; the theoretical probability
5 B B 1
is 36 ≈ .139. (7.1) 50. The empirical probability depends on the results .5 .7
2
of the simulation; the theoretical probability is 52 ≈ .038. (7.3) 51. False .6
A .4 .7 A .8
(7.2) 52. True (7.2) 53. False (7.3) 54. False (7.2) 55. True
.2 .2
(7.3) 56. False (7.2) 57. (A) .350 (B) 38 = .375 (C) 375 (7.1) .5 .3
C .1 C
58. - .0172; .0172; no (7.5) 59. 1 - 7C3 > 10C3 = 17 12
24 (7.1) 60. 51 ≈ .235
12
(7.3) 61. 51 ≈ .235 (7.3)
62. (A) .7 A B
57. .3 A B .1 A .3 .7
xi 2 3 4 5 6 .9
c d
B .9 .1
pi 9 12 10 4 1
36 36 36 36 36
A B C
(B) E1X2 = 10 3 (7.5) .1
B .3
A .6 .1 .3
63. E1X2 ≈ - $0.167; no; $ 110>32 ≈ $3.33 (7.5) 64. (A) 14 ; 1 to 3 .2
B C .2 .3 .5 S
(B) $3 (7.2, 7.4) 65. 1 - 1>74 ≈ .997 (7.2) 66. P1A B2 = P1B A2 if and 59. .6 A .5 61. .35 63. .212
C 0 0 1
only if P1A2 = P1B2 or P1AxB2 = 0. (7.3) 67. (A) .8 (B) .2
.3 C 1
P1A2 = .290; P1B2 = .290; P1A xB2 = .100; P1A B2 = .345; P1B A2 = .345
(C) .5 (7.2) 68. P1AxP2 = P1A2P1P A2 = .34 (7.3) 69. (A)
(B) No, since P1A x B2 ≠ P1A2P1B2 . (C) P1C2 = .880; P1D2 = .120;
A B C
65. S2 = 3.43 .35 .224; the probabilities of going from state A to states A,
P1CxD2 = 0; P1C D2 = 0; P1D C2 = 0 (D) Yes, since CxD = ∅; B, and C in two trials
dependent, since P1CxD2 = 0 and P1C2P1D2 ≠ 0 (7.3) 70. Plan A: A B C
E1X2 = $7.6 million; plan B: E1X2 = $7.8 million; plan B (7.5) 67. S3 = 3.212 .298 .494; the probabilities of going from state C to states
71. Payoff table: A, B, and C in three trials
xi $1,830 - $170 A B
A .4375 .5625 A B
pi .08 .92 69. n = 9 71. P 4 = c d ; S4 = 3.425 .5754
B .375 .625
E1X2 = - $10 (7.5)
72. 13C3 > 10C3 ≈ .580 (7.2) A B C
A .36 .16 .48 A B C
73. (A) 1 73. P 4 = B C .6 0 .4 S ; S4 = 3.452 .152 .3964
(B) E1X2 = (7.5) 74. .955 (7.4)
xi 0 1 2 2 C .4 .24 .36
24
75. ≈ .324 (7.4) A B C D
pi 12 9 1 74
22 22 22 A .0154 .3534 .0153 .6159

B 0 1 0 0
77. (A) D T (B) .6159 (C) .2962 (D) 0
C .0102 .2962 .0103 .6833
D 0 0 0 1
Chapter 8 81. (A) 3.25 .754 (B) 3.25 .754 (C) 3.25 .754 (D) 3.25 .754
Exercises 8.1 (E) The state matrices appear to approach the same matrix, S = 3.25 .754,
regardless of the values in the initial-state matrix.
16 .25 .75
1. c d 3. Not defined 5. 314 174 7. Not defined 83. Q = c d ; the rows of Q are the same as the matrix S from
14 .25 .75 R R′
9. S1 = 3.1 .94; S2 = 3.16 .844 11. S1 = 3.46 .544; Problem 81 85. (A) R = rain, R= = no rain R .4 .6
.6 (B) = c d
S2 = 3.376 .6244 13. S1 = 3.25 .754; S2 = 3.25 .754 R .06 .94
.4 R R .94
15. S1 = 3.5 .54; S2 = 3.65 .354 17. S1 = 3.71 .294; (C) Saturday: .196;
.06
S2 = 3.587 .4134 19. S1 = 3.65 .354; S2 = 3.605 .3954 Sunday: .12664
21. S1 = 3.7 .2 .14; S2 = 3.33 .35 .324 23. S1 = 3.35 .35 .34; 87. (A) .2 (B) X X′ (C) 32%; 39.2%
S2 = 3.465 .3 .2354 25. S1 = 3.53 .28 .194; .8 X X .8 X .8 .2
c d
S2 = 3.397 .325 .2784 27. S1 = 3.2 .3 .54; S2 = 3.27 .4 .334 .2 X = .2 .8
29. S1 = 3.28 .44 .284; S2 = 3.212 .492 .2964 89. (A) N = National Property, U = United Family, O = other companies
31. S1 = 3.24 .37 .394; S2 = 3.241 .446 .3134 U .85
.25
.3
.2 .3 .5 .1
33. .7 A B .9 35. P = C .1 .8 .1 S 37. Yes 39. No .65 N .05 .35
.1 .4 .2 .4 .15
.1
A B O .5
.6
41. No 43. Yes 45. .4 A B .3 A .4 .6 N U O
c d (B)
.7 B .7 .3 N .65 .25 .1 (C) 38.5%; 32% (D) 45%; 53.65%
U C .1 .85 .05 S
C .15 .35 .5
Answers A-25
91. (A) B = beginning agent, I = intermediate agent, (B) Year Data (%) Model (%) (C) 71.4%
T = terminated agent, Q = qualified agent
.4 1970 43.3 43.3
.5 B I .6 1980 51.5 51.2
1990 57.5 56.8
.1 .1 .3
2000 59.8 60.9
1 T Q 1
2010 58.5 63.9
59. GTT: 25%; NCJ: 25%; Dash: 50% 61. Poor: 20%; satisfactory: 40%;
B I T Q preferred: 40% 63. 51% 65. Stationary matrix = 3.25 .50 .254
(B) B .5 .4 .1 0 (C) .12; .3612 67. (A) 3.25 .754 (B) 42.5%; 51.25% (C) 60% rapid transit; 40% automobile
I 0 .6 .1 .3 69. (A) S1 = 3.334 .6664; S2 = 3.343 .6574; S3 = 3.347 .6534;
D T
T 0 0 1 0 S4 = 3.349 .6514
Q 0 0 0 1
(B) Year Data (%) Model (%) (C) 35%
HMO PPO FFS
1970 30.9 30.9
93. (A) HMO .8 .15 .05 (B) HMO: 34.75%; PPO: 37%; FFS: 28.25%
1980 33.3 33.4
PPO C .2 .7 .1 S (C) HMO: 42.2625%; PPO: 39.5875%;
FFS: 18.15%
1990 34.4 34.3
FFS .25 .3 .45
H R 2000 35.6 34.7
95. (A) H .847 .153 (B) 45.6% (C) 49.8% 2010 37.1 34.9
c d
R .174 .826
Exercises 8.3
Exercises 8.2 1. B, C 3. No absorbing states 5. A, D 7. B is an absorbing state; absorbing
1 0 0 0 0 0 chain 9. C is an absorbing state; not an absorbing chain 11. No 13. Yes
1 0 1 0
1. c d 3. c d 5. C 0 0 0S 7. C 0 0 0S 9. Yes B A C B D A C
0 0 0 1 15. Yes 17. No 19. Yes 21. B 1 0 0 23. B 1 0 0 0
0 0 1 0 0 0
A C .5 .2 .3 S D 0 1 0 0
11. No 13. No 15. Yes 17. No 19. Yes 21. No D T
C .1 .5 .4 A .4 .1 .3 .2
.4 .6 .375 .625
23. S = 3.4 .64; P = c d 25. S = 3.375 .6254; P = c d C .4 .3 0 .3
.4 .6 .375 .625
C A B B D A C
.3 .5 .2 25. C 1 0 0 27. B 1 0 0 0
27. S = 3.3 .5 .24; P = C .3 .5 .2 S A C .5 .2 .3 S D 0 1 0 0
D T
.3 .5 .2 B 0 1 0 A .2 .4 .1 .3
.6 .24 .16 C .2 .1 .5 .2
A B C
29. S = 3.6 .24 .164; P = C .6 .24 .16 S 31. SP = 3.2 .54; no,
A 1 0 0
.6 .24 .16 the sum of the entries
in S is not 1. 29. P = B C 0 1 0 S ; P1C to A2 = .2;
C .2 .8 0 P1C to B2 = .8.
33. SP = 30 04; no, the sum of the entries in S is not 1 35. False It will take an avg. of 2 trials to go from C to either A or B.
37. True 39. False 41. S = 3.3553 .64474
A B C
43. S = 3.3636 .4091 .22734 A 1 0 0
Red Blue 31. P = B C 1 0 0 S ; P1B to A2 = 1; P1C to A2 = 1. It will take an
45. (A) .6 (B) Red .4 .6
.4 Red Blue .8 c d C 1 0 0 avg. of 4 trials to go from B to A, and an avg. of
Blue .2 .8
.2 3 trials to go from C to A.
(C) 3.25 .754; in the long run, the red urn will be selected 25% of the time
and the blue urn 75% of the time. 47. (A) The state matrices alternate between A B C D
3.2 .84 and 3.8 .24; so they do not approach any one matrix. (B) The A 1 0 0 0
state matrices are all equal to S0, so S0 is a stationary matrix. (C) The powers B 0 1 0 0
33. P = D T ; P1C to A2 = .36; P1C to B2 = .64;
of P alternate between P and I (the 2 * 2 identity); so they do not approach a C .36 .64 0 0
P1D to A2 = .44; P1D to B2 = .56.
limiting matrix. (D) Parts (B) and (C) of Theorem 1 are not valid for this ma- D .44 .56 0 0
trix. Since P is not regular, this is not a contradiction. 49. (A) Since P is not It will take an avg. of 3.2 trials to go from C to either A or B, and an avg. of
regular, it may have more than one stationary matrix. (B) 3.5 0 .54 is an- 2.8 trials to go from D to either A or B. 35. (A) 3.2 .8 04
other stationary matrix. (C) P has an infinite number of stationary matrices. (B) 3.26 .74 04 37. (A) 31 0 04 (B) 31 0 04
1 0 0
39. (A) 3.44 .56 0 04 (B) 3.36 .64 0 04
51. P = C .25 0 .75 S ; each row of P is a stationary matrix for P.
(C) 3.408 .592 0 04 (D) 3.384 .616 0 04 41. False
0 0 1
53. (A) .39; .3; .284; .277 (B) Each entry of the second column of P k + 1 is A B C D
43. False 45. True 47. False 49. A 1 0 0 0
the product of a row of P and the second column of P k. Each entry of the latter
A B C D E B 0 1 0 0
is … Mk, so the product is … Mk. 55. 72.5% D T
51. A 1 0 0 0 0 C .6375 .3625 0 0
57. (A) S1 = 3.512 .4884; S2 = 3.568 .4324; S3 = 3.609 .3914;
B 0 1 0 0 0 D .7375 .2625 0 0
S4 = 3.639 .3614
C E .0875 .9125 0 0 0 U
D .1875 .8125 0 0 0
E .4375 .5625 0 0 0
A-26 Answers
A B C D A B C D
53. A 0 .52 0 .48 16. A 1 0 0 0 ; P1C to A2 = .2; P1C to B2 = .8; P1D to A2 = .3;
B 0 1 0 0 B 0 1 0 0 P1D to B2 = .7. It takes an avg. of 2 trials to go
D T D T
C 0 .36 0 .64 C .2 .8 0 0 from C to an absorbing state and an avg. of 3 trials
D 0 0 0 1 D .3 .7 0 0 to go from D to an absorbing state. (8.3)
59. (A) .370; .297; .227; .132; .045 (B) For large k, all entries of Qk are B D A C A B C A B C
close to 0. 61. (A) 75% (B) 12.5% (C) 7.5 months 63. (A) Company 21. B 1 0 0 0 (8.3) 22. (A) 3.1 .4 .54 (B) 3.1 .4 .54 (8.3)
A: 30%; company B: 15%; company C: 55% (B) 5 yr 65. (A) 91.52% D 0 1 0 0 A B C A B C
D T
(B) 4.96% (C) 6.32 days 67. (A) .375 (B) 1.75 exits A .1 .1 .6 .2 23. (A) 3.25 .75 04 (B) 3.55 .45 04 (8.3)
C .2 .2 .3 .3
24. No. Each row of P would contain a 0 and a 1, but none of the four
Chapter 8 Review Exercises matrices with this property is regular. (8.2)
A B A B
1. S1 = 3.32 .68]; S2 = 3.328 .6724. The probability of being in state 0 0 1
A after one trial is .32 and after two trials is .328; the probability of being in 25. Yes; for example, P = C 0 0 1 S is regular. (8.2)
state B after one trial is .68 and after two trials is .672. (8.1) .2 .3 .5
2. State A is absorbing; chain is absorbing. (8.2, 8.3) 3. No absorbing R B G
states; chain is regular. (8.2, 8.3) 4. No absorbing states; chain is neither. 26. (A) .25
Blue .6 (B) R .5 .25 .25 (C) Regular
(8.2, 8.3) 5. States B and C are absorbing; chain is absorbing. (8.2, 8.3) .2 B C .2 .6 .2 S
.5 Red .2 .3
6. States A and B are absorbing; chain is neither. (8.2, 8.3) G .6 .3 .1
.6
A B C .25 Green .1
A 0 1 0
7. B C .1 0 .9 S ; No absorbing states; chain is neither. (8.1, 8.2, 8.3) R B G
C 0 1 0 (D) R .4 .4 .2 In the long run, the red urn will be selected 40% of the
B C .4 .4 .2 S time, the blue urn 40% of the time, and the green urn
A B C G .4 .4 .2 20% of the time. (8.2)
A 0 1 0
R B G
8. B C .1 .2 .7 S ; C is absorbing; chain is absorbing. (8.1, 8.2, 8.3) 27. (A) (B) R 1 0 0 (C) Absorbing
C 0 0 1 Blue .6
B C .2 .6 .2 S
.2
G .6 .3 .1
A B C 1 Red .2 .3
A 0 0 1 .6
9. B C .1 .2 .7 S ; No absorbing states; chain is regular. (8.1, 8.2, 8.3) Green .1
C 0 1 0
R B G
(D) R 1 0 0 Once the red urn is selected, the blue and green urns will
A B C D BC1 0 0 S never be selected again. It will take an avg. of 3.67 trials
A .3 .2 0 .5 G 1 0 0 to reach the red urn from the blue urn and an avg. of 2.33
B 0 1 0 0 trials to reach the red urn from the green urn. (8.3)
10. D T ; B is absorbing; chain is neither. (8.1, 8.2, 8.3)
C 0 0 .2 .8 0 1 1 0
29. c d is one example (8.2) 30. c d is one example (8.3)
D 0 0 .3 .7 1 0 0 1
.3 .1 .6
31. C .3 .1 .6 S is one example (8.2)
B .3 .1 .6
.2
.8 A B C
11. .3 A .2 .3 A .3 .2 .5 12. (A) .3 1 0 0
.1 B C .8 0 .2 S (8.1) (B) .675 (8.1) 32. C 0 1 0 S is one example (8.3)
.5 .6 0 0 1
C C .1 .3 .6
33. If P is the transition matrix of an absorbing Markov chain with more than
A B one state, then P has a row with 1 on the main diagonal and 0’s elsewhere.
A B
A .25 .75 Every power of P has that same row, so no power of P has all positive entries,
13. S = c .25 .75]; P = c d (8.2)
B .25 .75 and the Markov chain is not regular. (8.2, 8.3) 34. If P is the transition ma-
trix of a regular Markov chain, then some power of P has all positive entries.
This is impossible for an absorbing Markov chain with more than one state
A B C
A .4 .48 .12 (see Problem 33), so the Markov chain is not absorbing. (8.2, 8.3)
A B C
14. S = 3.4 .48 .124; P = B £ .4 .48 .12 § (8.2) 35. SP = 3.3 .94; no, the sum of the entries in S is not 1. (8.1)
C .4 .48 .12 A B C D
A .392 .163 .134 .311
A B C B .392 .163 .134 .311
15. A 1 0 0 ; P1C to A2 = .75; P1C to B2 = .25. It takes an aver- 36. No limiting matrix (8.2, 8.3) 37. P = D T (8.2)
C .392 .163 .134 .311
BC 0 1 0 S age of 2.5 trials to go from C to an absorbing state. D .392 .163 .134 .311
C .75 .25 0 (8.3)
Answers A-27
X X X X 9. U.S. Postal Service Employees 11. Federal Income by Source, 2015

38. (A) .3 (B) X .7 .3 (C) 3.2 .84
.7 X X9 .5
c d
X′ .5 .5
Number of employees
1,000,000 Social
.5 insurance
800,000 Personal
X X income tax
taxes
600,000 (29%)
(D) 3.54 .464; 54% of the consumers will purchase brand X on the next purchase. (42%)
X X 400,000
(E) 3.625 .3754 (F) 62.5% (8.2) 39. (A) Brand A: 24%; brand B: 200,000
Corporate
32%; brand C: 44% (B) 4 yr (8.3) income tax
1990
1995
2000
2005
2010
2015
(9%)
40. (A) S1 = 3.48 .524; S2 = 3.66 .344; S3 = 3.76 .244 Other Borrowing to
Year (8%) cover deficit
(B) Year Data (%) Model (%) (C) 89% (8.2) (12%)
1995 14 14 13. (A)

2000 49 48 Relative
2005 68 66 Class Interval Tally Frequency Frequency
2010 79 76 30.5–34.5 1 .05
34.5–38.5 3 .15
41. (A) 63.75% (B) 15% (C) 8.75 yr (8.3) 38.5–42.5 7 .35
Red Pink White 42.5–46.5 6 .30

Red 1 0 0
46.5–50.5 2 .10
42. P = Pink C 1 0 0 S (8.3)
White 1 0 0 50.5–54.5 1 .05
43. (A) S1 = 3.244 .7564; S2 = 3.203 .7974; S3 = 3.174 .8264 (B) (C) .45; .2
8
(B) (C) 10.3% (8.2) (D)
Frequency
8
Year Data (%) Model (%)
1985 30.1 30.1
1995 24.7 24.4
30.5 38.5 46.5 54.5
2005 20.9 20.3 34.5 42.5 50.5
2010 19.3 17.4 30.5 54.5

0
15. (A)
Relative
Chapter 9 Class Interval Frequency Frequency
Exercises 9.1 -0.59 4.5 5 .05
1. (A) and (B) 4.5–9.5 54 .54
Class Interval Tally Frequency Relative Frequency 9.5–14.5 25 .25
14.5–19.5 13 .13
0.5–2.5 1 .1
19.5–24.5 0 .00
2.5–4.5 1 .1 24.5–29.5 1 .01
4.5–6.5 4 .4 29.5–34.5 2 .02
6.5–8.5 3 .3 100 1.00
8.5–10.5 1 .1 (B) Frequency Relative (C)

Relative frequency
frequency 50 .50
60 .60
Frequency
40 .40
Frequency Relative (C) The frequency tables and histograms 30 .30
frequency
4 0.4 are identical, but the data set in part 20 .20
10 .10
(B) is more spread out than that of
2 8 14 20 26
part (A). 0.5 9.5 19.5 29.5 Price–earnings ratios
4.5 14.5 24.5 34.5
Price–earnings ratios
0.5 2.5 4.5 6.5 8.5 10.5
(D)
3. (A) Let X min = 1.5, X max = 25.5, change Xscl from 1 to 2, and multiply Relative
Ymax and Yscl by 2; change Xscl from 1 to 4, and multiply Ymax and Class Cumulative Cumulative
Yscl by 4. Interval Frequency Frequency Frequency
(B) The shape becomes more symmetrical and more rectangular. -0.59 4.5 5 5 .05
5. Gross Domestic Product 7. China; Canada; United States 4.5–9.5 54 59 .59
9.5–14.5 25 84 .84
Billions of dollars
20
16 14.5–19.5 13 97 .97
12
8 19.5–24.5 0 97 .97
4 24.5–29.5 1 98 .98
x 29.5–34.5 2 100 1.00
1975
1985
1995
2005
2015
Year P1PE ratio between 4.5 and 14.52 = .79

A-28 Answers
(E) 17. Annual World Population Growth (D)

Cumulative frequency
Relative cumulative
100 100% 1
100
Relative
frequency
80 .80
80 Class Cumulative Cumulative
Millions
60 .60
60
40 .40
40
Interval Frequency Frequency Frequency
20 .20
20 1.95–2.15 21 21 .21
3 9 15 21 27 33 0
Price–earnings ratios 1900 1950 2000 2050 2.15–2.35 19 40 .40
Year
2.35–2.55 17 57 .57
2.55–2.75 14 71 .71
2.75–2.95 9 80 .80
19. Males, age 15–18 21. Calories
Females, age 15–18 Calories from fat 2.95–3.15 6 86 .86
Carbohydrate Burger
3.15–3.35 5 91 .91
3.35–3.55 4 95 .95
Extra patty 3.55–3.75 3 98 .98
Protein
Cheese 3.75–3.95 2 100 1.00
Fat P1GPA 7 2.952 = .2
Bacon
0 100 200 300 400 500 (E)
Cumulative frequency
Relative cumulative
RDA (g) Mayonnaise
100 1
frequency
80 .80
0 100 200 300 60 .60
40 .40
20 .20
23. Total Public School Enrollment by Region 25. 23 and 35;
2.15
2.55
2.95
3.35
3.75
1965 2020 the median age
2.35
2.75
3.15
3.55
3.95
decreased in the Grade-point averages
West
18% Northeast West Northeast 1950s and 1960s
15%
21% 26% but increased in
Midwest the other decades.
South 21%
Exercises 9.2
Midwest
33% South
28%
39%
1. 7 3. 0.6 5. 39 7. 1,980 9. Mean = 3; median = 3; mode = 3
11. Modal preference is chocolate. 13. Mean ≈ 4.4 15. The median
17. (A) Close to 3.5; close to 3.5 (B) Answer depends on results of simulation.
19. (A) 175, 175, 325, 525 (B) Let the four numbers be u, v, w, x, where u
and v are both equal to m3. Choose w so that the mean of w and m3 is m2; then
27. (A)
choose x so that the mean of u, v, w, and x is m1.
Class Relative 21. Mean ≈ 14.7; median = 11.5; mode = 10.1
Interval Frequency Frequency 23. Mean = 1,045.5 hr; median = 1,049.5 hr
1.95–2.15 21 .21 25. Mean = 55; median = 56; mode = 56
2.15–2.35 19 .19 27. Mean = 50.5 g; median = 50.55 g
2.35–2.55 17 .17 29. Mean = 2,560,700; median = 1,326,500; no mode 31. Median = 577
2.55–2.75 14 .14
2.75–2.95 9 .09 Exercises 9.3
2.95–3.15 6 .06 1. 26 3. 2 5. 32 7. 500 9. (A) x = 3.3; s ≈ 1.494 (B) 70%;
3.15–3.35 5 .05 100%; 100% (C) Yes (D) 11. 2.5 13. True
3
3.35–3.55 4 .04 15. False
17. True
3.55–3.75 3 .03
3.75–3.95 2 .02
100 1.00 1 2 3 4 5 6
19. (A) The first data set. It is more likely that the sum is close to 7, for exam-
(B) (C) ple, than to 2 or 12. (B) Answer depends on results of simulation.
Relative frequency
Relative frequency
25 .25 25 .25 21. x ≈ $4.35; s ≈ $2.45 23. x ≈ 8.7 hr; s ≈ 0.6 hr 25. x ≈ 5.1 min;
Frequency
Frequency
20 .20 20 .20 s ≈ 0.9 min 27. x ≈ 11.1; s ≈ 2.3

15 .15 15 .15
10 .10 10 .10
5 .05 5 .05 Exercises 9.4
5 32 1 5 1
1.95
2.35
2.75
3.15
3.55
3.95
2.05
2.45
2.85
3.25
3.65
1. ≈ .156 3. .276 5. ≈ .395 7. 9. 11.

32 81 16 16 16
2.15
2.55
2.95
3.35
3.75
2.25
2.65
3.05
3.45
3.85
Grade-point averages Grade-point averages

13. m = .75; s = .75 15. m = 1.333; s = .943
P(x) P(x)
0.5 .42 .42 0.5
.40
.30
.20
.14
.10
.02 .01
0 1 2 3 x 0 1 2 3 4 x
Answers A-29
17. m = 0; s = 0 19. .0008 21. .1157 23. .4823 57. (A) P1x2 = 5Cx 1.22 x 1.82 5 - x
P(x)
1.0 25. (A) .311 (B) .437 27. It (B) (C)
1.0 x P1x2 P(x) .410
is more likely that all answers are 0.4 .328
0 .328
wrong 1.1072 than that at least half
are right 1.0332.
1 .410 .205
2 .205 .051
0 1 2 3 4 x .006 .000
3 .051 0 1 2 3 4 5 x
4 .006
(D) m = 1; s = .89 59. (A) .0083
29. m = 2.4; s = 1.2 31. m = 2.4; s = 1.3 5 .000
P(x) P(x) (B) .0277 (C) .1861 (D) .9308
.30
0.4 0.3 .25
.31 .20
.28
.19 .14 Exercises 9.5
.14
.06 1. .4772 3. .3925 5. .4970 7. .6826 9. .4134 11. .3848 13. .5398
.0001
.05
.001
.05 .04 .01
.004
x 0 1 2 3 4 5 6 7 8 x 15. .7 17. 1.32 19. 2.59 21. .1700 23. .4927 25. .2454 27. .3413
0 1 2 3 4 5 6
29. .1974 31. .3085 33. .7734 35. False 37. True 39. True
41. No 43. Yes 45. No 47. Yes 49. Solve the inequality
33. (A) m = 17; s = 1.597 (B) .654 35. .238 np - 3 2npq Ú 0 to obtain n Ú 81. 51. .89 53. .16 55. .01 57. .01
37. The theoretical probability distribution is given by P1x2 = 3Cx 1.52 x 1.52 3 - x 59. 0.1 61. 0.2
= 3Cx 1.52 3
P(x)
0.4 .375 .375
.125 .125 10 40 0 40
0 0
0 1 2 3 x 63. (A) Approx. 82 (B) Answer depends on results of simulation. 65. 2.28%
67. 1.24% 69. .0031; either a rare event has happened or the company’s
Frequency of Heads in 100 Tosses of Three Coins claim is false. 71. 0.82% 73. .0158 75. 2.28% 77. A’s, 80.2 or
greater; B’s, 74.2–80.2; C’s, 65.8–74.2; D’s, 59.8–65.8; F’s, 59.8 or lower
Number Theoretical
of Heads Frequency Actual Frequency
Chapter 9 Review Exercises
0 12.5 (List your experimental results here)
1 37.5 1. (9.1)
Percentage of eligible voters
Percentage of eligible voters

2 37.5 64 64
3 12.5 60 60
56 56
39. p = .5 41. (A) m = 5; s = 1.581 (B) Answer depends on results of 52 52
simulation. 43. (A) .318 (B) .647 45. .0188 48 48
47. (A) P1x2 = 6Cx 1.052 x 1.952 6 - x

1996
2000
2004
2008
2012
2016
1996
2000
2004
2008
2012
2016
Year Year
x P1x2 (C) P(x)
(B) 1 2. (9.1) Graduate Less than 3. (A) .432
0 .735 .735 degree high school 0.4
12% 13% .288
1 .232 .216
.232
2 .031 .031 .002 .000 .000 .000 .064
3 .002 0 1 2 3 4 5 6 x Bachelor’s
degree 0 1 2 3
4 .000 (D) m = .30; s = .53 49. .998 19%
High school
diploma
5 .000 28%
Some college
6 .000 29%
(B) m = 1.2; s = .85 (9.4) 4. (A) x = 2.7 (B) 2.5 (C) 2

51. (A) .001 (B) .264 (C) .896 53. (A) P1x2 = 6Cx 1.62 x 1.42 6 - x (D) s = 1.34 (9.2, 9.3) 5. (A) 1.8 (B) .4641 (9.5)
(B) (C) P(x) 6. (A)
x P1x2 Class Relative
0.4
0 .004 .276
.311 Interval Frequency Frequency
1 .037 .187 9.5–11.5 1 .04
.138
2 .138 .037 .047 11.5–13.5 5 .20
.004
3 .276 0 1 2 3 4 5 6 x 13.5–15.5 12 .48
4 .311 (D) m = 3.6; s = 1.2 55. .000 864 15.5–17.5 6 .24
5 .187 17.5–19.5 1 .04
6 .047 25 1.00
A-30 Answers
(B) Frequency Relative (C) Appendix A
Relative frequency
frequency
15 0.6
Frequency
10 .4 Exercises A.1
5 .2 1. x = 5 3. x = 2 5. x = - 19 7. 4 … x 6 13
9. - 2 6 x 6 7
3
0 11. x … 4 13. 1 - 8, 24 15. 1 - ∞, 92 17. 1 - 7, - 54 19. x = -
10.5
12.5
14.5
16.5
18.5
2
15 3
9.5
11.5
13.5
15.5
17.5
19.5
21. y 6 - 23. u = - 25. x = 10 27. y Ú 3 29. x = 36
2 4
(D) Relative 36
31. m 6 33. 3 … x 6 7 or 33, 72 x
Class Cumulative Cumulative 7 3 7
Interval Frequency Frequency Frequency 35. - 20 … C … 20 or 3 - 20, 204 x

220 20
9.5–11.5 1 1 .04 3
37. y = x - 3 39. y = - 1A>B2x + 1C>B2 = 1 - Ax + C2 >B
11.5–13.5 5 6 .24 4
13.5–15.5 12 18 .72 5
41. C = 1F - 322 43. - 2 6 x … 1 or 1 - 2, 14
15.5–17.5 6 24 .96 9
17.5–19.5 1 25 1.00 x 45. Negative 47. 4,500 $35 tickets and
22 1
5,000 $55 tickets 49. Fund A: $180,000; Fund B: $320,000 51. $15,405
(E) 7. (A) x = 7 (B) s = 2.45
Cumulative frequncy
Relative cumulative
25 100% 1
53. (A) $420 (B) $55 55. 34 rounds 57. $32,000 59. 5,851 books
(C) 7.14 (9.2, 9.3)
61. (B) 6,180 books (C) At least $11.50 63. 5,000 65. 12.6 yr
frequncy
20 0.8
15 0.6 8. (A) P(x)
10 0.4 0.4
5 0.2
.313 Exercises A.2
.234 .234
0 1. (D) 3. (C) 5. y 7. y
11.5
13.5
15.5
17.5
19.5
.094 .094 5
(9.1) .016 .016 3
0 1 2 3 4 5 6 x
5 x 6 x
(B) m = 3; s = 1.22 (9.4) 9. m = 600; s = 15.49 (9.4)

10. (A) True (B) False (9.2, 9.3) 11. (A) False (B) True (C) True 5
(9.4, 9.5) 12. .999 (9.5) 13. (A) .9104 (B) .0668 (9.5) 9. Slope = 5; y int. = - 7 11. Slope = - ; y int. = - 9
2
14. (A) The first data set. Sums range from 2 to 12, but products range from 1 2
1 to 36. (B) Answer depend on results of simulation. (9.3) 13. Slope = ; y int. = 15. Slope = 2; x int. = - 5
4 3
15. (A) x = 14.6; s = 1.83 (B) x = 14.6; s = 1.78; (9.2, 9.3) 6
17. Slope = 8; x int. = 5 19. Slope = ; x int. = - 7
16. (A) .0322 (B) .0355 (9.4) 17. .421 (9.4) 18. (A) 10, 10, 20, 7
20, 90, 90, 90, 90, 90, 90 (B) No (9.2) 19. (A) .179 (B) The normal 1 1
21. y = 2x + 1 23. y = - x + 6 25. x int.: ; y int.: 1; y = - 2x + 1
distribution is continuous, not discrete, so the correct analogue of part (A) is 3 2
x y
P17.5 … x … 8.52 ≈ .18 (using Appendix C). (9.5) 27. x int.: - 3; y int.:1; y = + 1 29.
3 5
20. (A) m = 10.8; s = 1.039 (B) .282 (C) Answer depends on results
of simulation. (9.4) 21. (A) x = 10 (B) 10 (C) 5 (D) s = 5.14 (9.2,
5 x
9.3) 22. Modal preference is soft drink (9.2)
23. (A) Class Relative

Interval Frequency Frequency 3
31. y 33. y 35. - 4 37. -
5 5 5
29.5–31.5 3 .086 x53
31.5–33.5 7 .2
5 x 5 x
33.5–35.5 14 .4 y 5 22
35.5–37.5 7 .2
39. 2 41. (A) 43. y
37.5–39.5 4 .114 (C) y 3,000
35 1.00 10
(B)
10 x
(B) Frequency (C) x = 34.61; s = 2.22
and relative (9.1, 9.2, 9.3) 100 x
frequency
Frequency Relative histogram 24. (A) 57.62% 45. (A) y (B) x int.: 3.5; y int.: - 4.2
frequency
15 0.4 Frequency (B) 6.68% (9.5) 10
and relative 25. (A) m = 140; s = 6.48
frequency
polygon (B) Yes (C) .939 10 x
(D) .0125
(E) 0.07
29.5
31.5
33.5
35.5
37.5
39.5
10
30.5
32.5
34.5
36.5
38.5
(C) (D) x int.: 3.5; y int.: - 4.2
210 10
120 160
0
26. .999 (9.4) (9.4, 9.5) 210
Answers A-31
47. x = 4, y = - 3 49. x = - 1.5, y = - 3.5 51. y = 5x - 15 (B) 2025 13. (A) y

(B) 1,050,000 (C) 1,359,000
Licensed Drivers
1 20 1
53. y = - 2x + 7 55. y = x - 57. y = - 3.2x + 30.86
(millions)
3 3
2 2 11 0.5
59. (A) m = (B) - 2x + 3y = 11 (C) y = x +
3 3 3
5 5 7 1 x
61. (A) m = - (B) 5x + 4y = - 14 (C) y = - x - 0 0.5
4 4 2 Population (millions)
63. (A) Not defined (B) x = 5 (C) None 65. (A) m = 0
15. (A) S (B) $662 billion
(B) y = 5 (C) y = 5 67. The graphs have the same y int., (0, 2).
Net Sales (millions)

600
69. C = 124 + 0.12x; 1,050 donuts 71. (A) C = 75x + 1,647
(B) C (C) The y int., $1,647, is the fixed cost and the 300
slope, $75, is the cost per club.
15,000
73. (A) R = 1.4C - 7 (B) $137 0 10 20 t
10,000
75. (A) V = - 7,500t + 157,000 Years since 2000
5,000
17. (A) y (B) 2°F (C) 22.75%
0 50 100 150 x
Ethylene Glycol
80
(B) $112,000 (C) During the 12th year (D) V
(%wt)
$120,000 (6, 112,000)
$80,000 (11.6, 70,000) 24
$40,000 260 230 0 30 x

Degrees (8 F)
0 10 20 t
Years 19. (A) The rate of change of height with respect to Dbh is 1.37 ft/in.
77. (A) T = - 1.84x + 212 (B) 205.56°F (C) 6,522 ft (B) Height increases by approximately 1.37 ft. (C) 18 ft (D) 20 in.
21. (A) Undergraduate male enrollment is increasing at a rate of 87,000
(D) T 79. (A) T = 70 - 3.6A (B) 10,000 ft
students per year; undergraduate female enrollment is increasing
Temperature (8F)
220
210 (3.5, 205.56) at a rate of 140,000 students per year. (B) Male: 8.6 million;
200 (6.522, 200)
female: 11.5 million (C) 2026 23. y = 0.061x + 50.703; 54.67°F
190
180 25. Men: y = - 0.070x + 49.058; women: y = - 0.085x + 54.858; yes
0 10 20 x 27. Supply: y = 0.2x + 0.87; demand: y = - 0.15x + 3.5; equilibrium
Thousands of Feet
price = $2.37
81. (A) N = - 0.0063t + 2.76 (B) 2.45 persons 83. (A) f = - 0.49t + 21
(B) 2028 85. (A) p = 0.001x + 5.4 (B) p = - 0.001x + 13
(C) (3,800, 9.2) (D) p Equilibrium point 87. (A) s =

2
5
w Appendix B
Supply
10
Exercises B.1
Dollars
9.5
1. vu 3. 13 + 72 + y 5. u + v 7. T 9. T 11. F 13. T 15. T
9 Demand 17. T 19. T 21. F 23. T 25. T 27. No 29. (A) F (B) T (C) T
0 3,400 4,200 x 31. 22 and p are two examples of infinitely many. 33. (A) N, Z, Q, R (B) R
(B) 8 in. (C) 9 lb (C) Q, R (D) Q, R 35. (A) F, since, for example, 213 - 12 ≠ 2 # 3 - 1
(B) F, since, for example, 18 - 42 - 2 ≠ 8 - 14 - 22 (C) T (D) F,
1
since, for example, 18 , 42 , 2 ≠ 8 , 14 , 22. 37.
11
Exercises A.3 39. (A) 2.166 666 666 . . . (B) 4.582 575 69 . . . (C) 0.437 500 000 . . .
1. (A) w = 49 + 1.7h (B) The rate of change of weight with respect to (D) 0.261 261 261 . . . 41. (A) 3 (B) 2 43. (A) 2 (B) 6 45. $16.42
height is 1.7 kg/in. (C) 55.8 kg (D) 5′6.5″ 47. 2.8%
3. (A) P = 0.445d + 14.7 (B) The rate of change of pressure with respect
to depth is 0.445lb/in.2 per ft. (C) 37 lb/in.2 (D) 99 ft Exercises B.2
5. (A) a = 2,880 - 24t (B) - 24 ft/sec (C) 24 ft/sec 1. 3 3. x3 + 4x2 - 2x + 5 5. x3 + 1
7. s = 0.6t + 331; the rate of change of the speed of sound with respect to 7. 2x5 + 3x4 - 2x3 + 11x2 - 5x + 6 9. - 5u + 2 11. 6a2 + 6a
temperature is 0.6 m/s per °C. 9. (A) y 13. a2 - b2 15. 6x2 - 7x - 5 17. 2x2 + xy - 6y 2 19. 9y 2 - 4
21. - 4x2 + 12x - 9 23. 16m2 - 9n2 25. 9u2 + 24uv + 16v2
Production (%)
U.S. Fossil Fuel
100
80 27. a3 - b3 29. x2 - 2xy + y 2 - 9z2 31. 1 33. x4 - 2x2y 2 + y 4
60
40 35. - 40ab 37. - 4m + 8 39. - 6xy 41. u3 + 3u2v + 3uv2 + v3
20 43. x3 - 6x2y + 12xy 2 - 8y 3 45. 2x2 - 2xy + 3y 2
0 5 10 15 20 25 30 x 47. x4 - 10x3 + 27x2 - 10x + 1 49. 4x3 - 14x2 + 8x - 6 51. m + n
Years since 1985
53. No change 55. 11 + 12 2 ≠ 12 + 12; either a or b must be 0
(B) The rate of change of fossil fuel production is - 0.19% per year. 57. 0.09x + 0.12110,000 - x2 = 1,200 - 0.03x
(C) 76% of total production (D) 2058 11. (A) f 59. 20x + 3013x2 + 5014,000 - x - 3x2 = 200,000 - 90x
Female Smokers (%)
25 61. 0.02x + 0.06110 - x2 = 0.6 - 0.04x

20
15
10 Exercises B.3
5
0 3 6 9 12 15 18 t 1. 3m2 12m2 - 3m - 12 3. 2uv14u2 - 3uv + 2v2 2
Years since 1997 5. 17m + 5212m - 32 7. 14ab - 1212c + d2 9. 12x - 121x + 22
A-32 Answers
11. 1y - 1213y + 22 13. 1x + 4212x - 12 15. 1w + x21y - z2 Exercises B.6

17. 1a - 3b21m + 2n2 19. 13y + 221y - 12 21. 1u - 5v21u + 3v2
5 3 5
23. Not factorable 25. 1wx - y21wx + y2 27. 13m - n2 2 1. 6 2 x 3. 2 132x2y 3 2 3 5. 2x2 + y 2 1not x + y2 7. 5x3>4
29. Not factorable 31. 41z - 321z - 42 33. 2x2 1x - 221x - 102 9. 1 2x y 2
2 3>5
11. x1>3 + y 1>3 13. 5 15. 64 17. - 7 19. - 16
35. x12y - 32 2 37. 12m - 3n213m + 4n2 39. uv12u - v212u + v2 8 1
21. 23. 25. x2>5 27. m 29. 2x>y 2 31. xy 2 >2
41. 2x1x2 - x + 42 43. 12x - 3y214x2 + 6xy + 9y 2 2 125 27
45. xy1x + 221x2 - 2x + 42 47. 3 1x + 22 - 3y4 3 1x + 22 + 3y4 33. 1> 124x7>12 2 35. 2x + 3 37. 30x5 13x 39. 2 41. 12x - 6x35>4
49. Not factorable 51. 16x - 6y - 121x - y + 42 6m1>2 6m 1
43. 3u - 13u1>2v1>2 + 4v 45. 36m3>2 - 1>2 + 1>2 -
53. 1y - 221y + 221y 2 + 12 55. 31x - y2 2 15xy - 5y 2 + 4x2 n n n
1 2 1
57. True 59. False 47. 9x - 6x1>2y 1>2 + y 49. x1>3 + x -1>3 51. x -1>4 + x -2>3
2 3 3
1 -1>6 1 21x + 32 1x - 2
Exercises B.4 53. x - 55. 4n 13mn 57.
2 4 x - 2
15x2 + 10x - 6 15m2 + 14m - 6 1 1
1. 39/7 3. 495 5. 8d 6 7. 9. 59. 71x - y21 1x + 1y2 61. 63.
180 36m3 xy 15xy 2x + h + 1x
1 x - 6 - 3x - 9 2 1
11. 13. 15. 17. 65. 67. x = y = 1 is one of many choices.
x1x - 42 x1x - 32 1x - 221x + 12 2 x - 1 1t + x21 1t + 1x2
5 x2 + 8x - 16 7x2 - 2x - 3 x1y - x2 69. x = y = 1 is one of many choices. 71. False 73. False 75. False
19. 21. 23. 25.
a - 1 x1x - 421x + 42 61x + 12 2 y12x - y2 x + 8 x - 2
77. True 79. True 81. False 83. 85.
- 17c + 16 1 -1 x - y 21x + 32 3>2
21x - 12 3>2
27. 29. 31. 33.
151c - 12 x - 3 2x1x + h2 x + y x + 6
87. 89. 103.2 91. 0.0805 93. 4,588
35. (A) Incorrect (B) x + 1 37. (A) Incorrect (B) 2x + h 31x + 22 5>3
95. (A) and (E); (B) and (F); (C) and (D)
x2 - x - 3 - 2x - h
39. (A) Incorrect (B) 41. (A) Correct 43.
x + 1 2 2
31x + h2 x Exercises B.7
x1x - 32 4
45.
x - 1 1. { 111 3. - , 2 5. - 2, 6 7. 0, 2 9. 3 { 2 13 11. - 2 { 12
3
15 3 1
13. 0, 15. { 17. , - 3 19. 1 - 1 { 152 >2 21. 13 { 132 >2
Exercises B.5 2 2 2
1
23. No real solution 25. 1 - 3 { 1112 >2 27. { 13 29. - , 2
1. 2>x9 3. 3w7 >2 5. 2>x3 7. 1>w5 9. 4>a6 11. 1>a6 13. 1>8x12 2
31. 1x - 221x + 422 33. Not factorable in the integers
15. 8.23 * 1010 17. 7.83 * 10 -1 19. 3.4 * 10 -5 21. 40,000
35. 12x - 921x + 122 37. 14x - 721x + 622 39. r = 1A>P - 1
23. 0.007 25. 61,710,000 27. 0.000 808 29. 1 31. 1014 33. y 6 >25x4
41. If c 6 4, there are two distinct real roots; if c = 4, there is one real
7 1 5 3
35. 4x6 >25 37. 4y 3 >3x5 39. - x -3 41. x2 - + 4x -2 double root; and if c 7 4, there are no real roots. 43. - 2 45. { 110
4 4 2 2
2
x 1x - 32 21x - 12 47. { 13, { 15 49. 1,575 bottles at $4 each 51. 13.64% 53. 8 ft/sec;
43. 45. 47. 2.4 * 1010; 24,000,000,000 4 12 or 5.66 ft/sec
1x - 12 3 x3
bc1c + b2
49. 3.125 * 104; 31,250 51. 64 55. uv 57. 2
c + bc + b2
59. (A) $60,598 (B) $1,341 (C) 2.21% 61. (A) 9 * 10 -6 (B) 0.000 009
(C) 0.0009% 63. 1,194,000
INDEX
NOTE: Page numbers preceded by G refer to online Chapter 10: Games and Decisions. Page numbers preceded by A refer to online
Appendix C: Special Topics.
A Big M method of linear programming

for maximization problem with mixed problem
Abscissa of points, 541
constraints, 308–12
Absolute value function, 35–6
minimization by, 316–18
Absorbing Markov chains, 460–2
overview, 312–16
Absorbing states, in Markov chains, 459–60
summary of, 318
ac test for factorability, 581
Binomial theorem, A13–A16
Acceptable probability assignment, 377
Boole, George, 329
Addition
Boundary lines of half-planes, 236–7
counting principles for, 346–9
Bounded functions, 69
of fractions, 587–8
Bounded solution regions, 246
of polynomials, 575–6
Broken-line graphs, 580–1, 585
of real numbers, 568
Butler Tactical Parachute Systems,
Additive inverse of real numbers, 203
556–7
Algebra, 567–609
integer exponents, 591–2
C
polynomials, factoring, 579–84
polynomials, operations on, 573–9 Canceling, in fraction operations, 586
quadratic equations, 601–9 Capital One Bank (VA), 101
rational exponents and radicals, 595–601 Cardano, Girolamo, 372
rational expressions, operations on, 585–90 Cartesian coordinate system, 541
real numbers, 567–73 Cayley, Arthur, 192
scientific notation, 592–5 Central tendency, measures of
Alternating series, A4 mean, 490–3
Amortization, 137–42 median, 493–5
Amount (future value) from investment, 80 mode, 495–7
Analytic geometry, fundamental Change, rate of, 555–6
theorem of, 541 Class frequency, 482
Annual nominal rate, 121–2 Class intervals, for frequency
Annual percentage yield (APY), 120–2 tables, 482
Annuities Closed intervals, 534
future value of, 126–9 Coefficient matrix, 168
present value of, 134–7 Coefficients, 574–5
Approximate empirical probability, 380 Column matrices, 167
Arithmetic and geometric sequences, Combinations, 359–61
A7–A12 Common factors, 579–80
Arithmetic mean, A5 Common logarithms, 89
Artificial variables, 309, 312 Commutative property, 195
Associative property of real numbers, 568–9 Commutative property of real
Asymptotes numbers, 568–9
of rational functions, 65–6, 68–9 Complement of events, 390–2
Atmospheres of pressure, 562 Complement of sets, 342
Augmented matrices, 168–73, 169–73, 310 Completing the square
process of, 50
B to solve quadratic equations, 604–5
Bar graphs, 479–8 Compound events, 374
Base of exponential functions, 74 Compound fractions, 588–9
Basic and nonbasic variables in linear Compound interest
programming, 273–4, 279–80 calculating, 112–15
Bayes’ formula, 413–18 continuous, 115–17
Best fit line, 558 Compound propositions, in logic, 332
I-1
I-2 Index
Conceptual insight Conditional propositions, in logic, 336–7

absolute value, 36 Conjunction, in logic, 331
amortization, 139 Connectives, in logic, 330–3
canceling, 586 Consistent systems of linear equations, 155
coefficients, A15 Constant functions, 25
compound interest, 114 Constant matrix, 168
conditional, in logic, 332 Constant profit line, 252–3
continuous compound interest, 117 Contingency, in logic, 334
cumulative frequency polygons, 486 Continuous compound interest, 81, 115–17
DeMorgan’s laws, 336 Contradiction, in logic, 334
expected values of games, G11 Contrapositive, in logic, 332–3, 336–7
exponential functions, 75 Converse, in logic, 332–3
finite geometric series, 127, 135 Coordinate axes, 541
graphing calculator approximations, 454 Coordinates, 267, 541
graph transformation properties, 51 Corner points, 245, 255, 268, 274
independent events, 407 Costs, 28
infinity symbol (∞), 116 Counting principles
integer programming, 258 addition, 346–9
interval notation, 535 combinations, 359–61
inverses of matrices, 217 factorials, 355–6
limiting matrix of Markov chains, 466 multiplication, 349–52
linear programming for games, permutations, 356–8
G22, G26 Cube function, 35–6
logarithmic functions, 86 Cube root function, 35–6
logarithms to solve equations, 119 Cumulative frequency polygons, 585–6
matrix multiplication, 195–6, 198 Curve fitting, regression analysis as, 557
means of grouped data and ungrouped data, 493
multiplication in simplex method, 290 D
multiplication of problem constraints in dual problems, 300 Dantzig, George B., 265
negation, 569 Data description, 478–529
notation for functions, 27 bar graphs for, 479–80
optimal solutions in linear programming, 256 broken-line graphs for, 480–1
permutations and combinations, 361 frequency distributions for, 482–3
piecewise-defined functions, 109 frequency polygons and cumulative frequency polygons for,
pivot operation, 282 485–6
population standard deviation, 502 histograms for, 483–5
probability distributions, 426 mean, as measure of central tendency for, 490–3
quadratic equations, 602 median, as measure of central tendency for, 493–5
quadratic functions, 47 mode, as measure of central tendency for, 495–7
rational exponents, 597 pie graphs for, 481
rational functions, 67–8 range, as measure of dispersion for, 500
recessive rows and columns, in game theory, G15 standard deviation and, 500–4
sample spaces, 375 statistics for, 483
sets, 340 Data range, for frequency tables, 482
simple interest as linear function, 107 Decibel scale, logarithmic, 84
simple versus compound interest, 122 Decision making and expected
slope of lines, 545 value, 426–7
solution regions for systems, 244 Decision variables, in linear
technology matrix, 224 programming, 254
transition matrix, 439 Degree of the term, in a polynomial, 574
truth tables, 335 DeMorgan’s laws, 336
tuberculosis screening, 416 Denominators, rationalizing, 599
union and intersection of events, 388 Dependent systems of linear equations, 155, 182–3
Venn diagrams, 349 Dependent variables, 23
zero of functions, 47–8 Descent, rate of, 556–7
Conditional, in logic, 331 Dimensions of matrices, 167
Conditional probability, 399–402 Disjunction, in logic, 331
Index I-3
Dispersion measures augmented matrices, 171–2

range, 500 bar graphs, 480
standard deviation of grouped data, 503–4 bounded and unbounded solution regions, 246
standard deviation of ungrouped data, 500–3 bounded functions, 69
Distributive properties of real numbers, 568–9, 574 broken-line graphs, 481
Division of fractions, 586–7 complementary events, 391
Division properties of real numbers, 570 compound interest, 80
Domains of functions, 22, 25, 27, 74 constant-profit line, 253–4
Dominant rows and columns, in game theory, G14–G15 correspondence, 22
Double inequalities, 534, 536 dual problems, 298
Double subscript notation, for matrix position, 167 equally likely assumptions, 382
Dual problem, in linear programming, 294–6 equivalent equations, 532
for mn games, G25–G30 expected values, 424
exponential functions with base e, 77
E future value, 129
Elementary functions, 34–6 graphing feasible regions, 287
Elimination by addition method for solving systems of linear identity matrices, 203
equations, 157–9 independent events, 408
Empirical probability, 380–2 inequalities, 534
Encoding matrices, 210 inequalities, graphs of, 535
Endpoints of intervals, 534–5 inverse of square matrices, 208–9
Equally likely assumptions, 380–3 investment analysis, 219
Equations Leontief input-output analysis, 226
functions specified by, 23–5 linear inequalities, 238
quadratic, 601–9 linear programming problem components, 295
quadratic functions, inequalities and, 46–50 linear regression, 559
quartic, 607–8 LOGIC menu on calculators, 334
solver for, on graphing calculators, 120 mathematical modeling, 184
in two variables, 20–1 mean, median, and mode, 497
Equilibrium point, 551, 607 mn matrix games, G30
Equilibrium price, 551 mortgages, 138, 141
Equilibrium quantity, 551 natural and common logarithms, 91
Equity, formula for, 139–41 nonstrictly determined matrix
Equivalences, in logic, 335–7 games, G16
Equivalent equations, 531–2 one-to-one functions, 85
Equivalent systems of linear equations, 155 optimal solutions in linear
e -system of linear equations, 267, 269–70 programming, 256
Events permutations and combinations, 361
complement of, 390–2 plotting equation solutions, 542
compound, 374 point-by-point plotting, 21
independent, 405–9 probability trees, 405, 415
mutually exclusive, 408 quadratic functions, 47, 50
probability of, 377–80 real number multiplication versus matrix multiplication, 196
product rule in intersection of, 402–3 reduced matrices, 181
sample spaces and, 373–6 reflections, stretches, and shrinks, 39–40
simple, 373–4, 377 Roth IRAs (individual retirement accounts), 131
union and intersection of, 387–90 saddle values, G4, G14
Expanded coordinates, 267 sets, 340
Expected value simple interest, 106
decision making and, 426–7 simple versus compound interest, 122
of games, G9–G12 slack and surplus variables, 321
means of grouped data and ungrouped data as, 493 slopes of lines, 556
of random variables, 422–6 standard maximization problems, 266
Experiments, random, 373 state matrices of Markov chains, 460–1
Explore and discuss stationary matrices of Markov
absorbing states of Markov chains, 441 chains, 449
annual percentage yield, 120 strictly determined matrix games, G24
I-4 Index
Explore and discuss (Continued) Functions

substitution to solve systems of linear definition of, 21–2
equations, 157 elementary, 34–6
systems of linear equations, 155 equations in two variables, 20–1
table methods, 271, 273 equations to specify, 23–5
transition probability matrix, 437 graphs of quadratic, 50–3
truth tables, 337 notation for, 25–8
Venn diagrams, 343, 349 piecewise-defined, 41–2
Exponential functions polynomial, 62–4
equation solving, 90–1 probability, 376
logarithmic–exponential quadratic functions, equations, and inequalities, 46–50
conversions, 86–7 rational, 65–9
Exponents reflections, stretches, and shrinks, 38–40
integer, 591–2 regression polynomials, 64–5
natural number, 573 vertical and horizontal shifts, 36–8
rational, 595–601 Fundamental principle of duality, 296
Extrapolation, 559 Fundamental property of fractions, 585
Fundamental sample space, 375
F Fundamental theorem of analytic
Factorials, A14 geometry, 541
Factoring Fundamental theorem of game theory, G12
polynomials, 579–84 Fundamental theorem of linear programming
quadratic equations solved by, 602–3 version 1, 254
quadratic formula and, 606–7 version 2, 269
Feasible region for systems, 244, 254 version 3, 278
Feasible solutions, 267, 274 Future value, 105, 126–9
Federal Bureau of Investigation (FBI), 102
Fermat, Pierre de, 372 G
Finance, mathematics of Games, G1–G33. See also Matrices
amortization, 137–42 expected value of, G9–G11
annual percentage yield, 120–2 fundamental theorem of game theory, G12
compound interest, 112–15 linear programming and mn games, simplex method and dual
continuous compound interest, 115–17 problem, G25–G30
future value of annuities, 126–9 nonstrictly determined, example of, G8–G9
growth and time, 117–20 nonstrictly determined matrix, G5–G6
interest rates, approximating, 131–2 pure and mixed strategies for, G9
present value of annuities, 134–7 recessive rows and columns in, G14–G16
problem-solving strategy for, 142–3 solution to 22 matrix, G12–G14
simple interest, 105–9 strictly determined matrix, G2–G5
sinking funds, 129–31 Games and Economic Behavior (von Neumann and
Finance, mathematics of, 104–51 Morgenstern), G1
Finite geometric series, 127, 135 Gauss, Carl Friedrich, 177
Finite sequences, A2 Gauss–Jordan elimination, 173,
Finite series, A3, A9–A11 177–82, 271
Finite sets, 339 General term of a sequence, A2–A3
First-degree equations. See Linear equations Geometric interpretation of simplex method, 284–5
First property of exponents, 573 Geometric sequences, A7–A12
First-state matrix, 438 Graphing calculators
Fixed costs, 28 equation solver on, 120
Fraction properties of real numbers, 571 histograms construction on, 584–5
Fractional expressions, 585 intercepts found on, 543–4
Fractions, LCD (least common denominator) for Markov chains approximations, 453–4,
of, 532 468–9
Fractions, operations on, 585–9, 592 matrices on, 167, 170, 189
Frequency distributions, 582–3, 500 simple versus compound interest on, 122
Frequency of events, 380 standard deviation on, 502
Frequency polygons, 485–6 systems of linear equations solved on, 156
Index I-5
Graphs Initial basic feasible solution, 278

bar, 479–80 Initial simplex tableau, 279–80, 310
broken-line, 480–1, 485 Initial-state distribution matrix, 438
of exponential functions, 75–6 Initial-state probability matrix, 438
line, 535–6 Input–output analysis. See Leontief input–output
of linear equations in two variables, 541–4 analysis
pie, 481 i-system of linear inequalities, 267, 269–70
of piecewise-defined functions, 41–2 Integer exponents, 591–2
point-by-point plotting, 20 Integer programming, 258
quadratic equation properties and, 52 Interest
of quadratic functions, 50–3 on amortized debt, 137–8
of rational functions, 66–7 approximating rates of, 131–2
to solve systems of linear inequalities, 244–6 compound, 79–81, 112–15
stretches, shrinks, and reflections of, 38–40 continuous compound, 115–17
of systems of linear equations, 153–6 simple, 105–9
transformation of, 36–8 Interpolation, 559
Grouped data Intersection
mean of, 491–3 of events, 387–90
median of, 494–5 of sets, 341–2
standard deviation of, 503–4 Interval notation, 534–5
Grouping, factoring by, 580 Invariant optimal strategies, in game theory, G19
Growth time, for investments, 117–20 Inverse
of matrices, 215–17
H of square matrices, 203–9
Half-life of radioactive material, 78 Inverse property of real numbers, 568–9
Half-planes, 236 Irrational numbers, 76
Harsanyi, John, G1
Histograms, 421–2
J
for data description, 483–5 Jack Haggerty Forest, Lakehead University, Canada, 561
of probability distributions, 421–2 Jordan, Wilhelm, 177
Horizontal asymptotes
of rational functions, 65–6, 68–9 L
Horizontal (x) axis, of coordinate system, 541 Lakehead University, Canada, 561
Horizontal lines, 542, 544 Law of large numbers, 502
Horizontally translating graphs, 37–8 Leading coefficients, 63
Horizontal and vertical shifts, 36–8 Least common denominator (LCD) of fractions,
532, 587
I Left half-planes, 236
Identity function, 35–6 Leftmost variables, in reduced matrices, 181
Identity matrices, 462 Leontief, Wassily, 222
Identity property of real numbers, 568–9 Leontief input-output analysis
Implication, in logic, 335–7 overview, 222
Inconsistent systems of linear equations, 155 two-industry model, 223–5
Independent events, 405–9 three-industry model, 226–8
Independent systems of linear equations, 155 Like terms, in polynomials, 574–5
Independent variables, 23 Limiting matrix, in Markov chains, 462–8
Index of radicals, 596 Line of symmetry, in parabolas of quadratic equations, 51
Inequalities Linear equation systems, 152–234
double, 534, 536 augmented matrices to solve, 169–73
first-degree, 531 elimination by addition method to
linear equations and, 533–5 solve, 157–9
quadratic functions, equations and, 46–50 Gauss–Jordan elimination to solve, 173, 177–82
slack variables for problem constraint, 267 graphing, 153–6
symbols for (6, 7), 533 matrix equations and, 214–17
Inferential statistical process, 483 reduced matrices for, 176–7
Infinite series, A11 substitution method to solve, 156–7
Infinite sets, 339 in two variables, 153
I-6 Index
Linear equation systems, Leontief input–output analysis absorbing chains, 460–1

two-industry model, 222–5 absorbing states, 459–60
three-industry model, 226–8 graphing calculator approximations of, 453–4, 468–9
Linear equation systems, matrices introduction to, 437–9
addition and subtraction of, 189–90 limiting matrix, 462–8
identity, 202–3 powers of transition matrices in, 441–2
inverse of square, 203–9 regular, 449–51
multiplication by a number, 190–2 standard form of absorbing, 461–2
multiplication of, 192–6 stationary matrices in, 448–9
overview of, 166–8 transition and state matrices in, 439–41
Linear equations, 530–69 Mathematical models, 251–3, 555
on Cartesian coordinate system, 541 Matrices
double inequalities and, 536 addition and subtraction of, 189–90
inequalities and, 533–5 augmented, 169–73
line graphs of, 535–6 identity, 202–3
linear regression, 557–61 inverse of square, 203–9
point-slope form of, 547–8 linear equations and equations of, 214–17
slope as rate of change, 555–7 multiplication of, 192–6
slope of a line, 544–6 multiplication by a number, 190–2
slope-intercept form of, 546–7 overview of, 166–8
solving, 531–3 reduced, 176–7
in two variables, graph of, 541–4 Maximization problems
Linear inequalities linear programming solution for,
graphically solving systems of, 244–6 252–3, 266
in two variables, 236–40 with mixed problem constraints, 308–12
Linear programming Mean, as measure of central tendency, A5
description of, 254 Median, as measure of central tendency, 493–5
example application of, 250–4 Méré, Chevalier de, 372
geometric method for, 255–6 Mixed and pure strategies for games, G9
Linear programming, simplex method for, 265–328 Minimization problems, linear programming solution for, 296–301
basic and nonbasic variables in, 273–4 Mode, as measure of central tendency, 495–7
big M method, 312–18 Morgenstern, Oscar, G1
dual problem formation, 294–6 mn matrix, 167
example of, 277–9 mn matrix games, G2, G25–G30
geometric interpretation of, 284–5 Multiplication
initial simplex tableau in, 279–80 counting principles for, 349–52
maximization problem with mixed problem of fractions, 586–7
constraints, 308–12 of polynomials, 576–7
minimization problem solution, 296–301 Multiplicative inverse of real numbers,
for mn games, with dual problem, 203, 569
G25–G30 Mutually exclusive events, 408
pivot operation in, 280–4
slack variables for problem constraint inequalities, 267
N
standard maximization problems, 266 Nash, John, G1
summary of, 285–8 Natural logarithms, 89
table method as introduction to, 267–72 Natural number exponents, 573
Linear relationship, 555 Negation, in logic, 330
Logarithmic regression, 92 Negative numbers, multiplying and
Logic dividing by, 534
implications and equivalences, 335–7 Negative properties of real numbers, 570
propositions and connectives, 330–3 Negative real numbers, 568
truth tables, 333–4 Neumann, John von, G1
Lower half-planes, 236 New York Stock Exchange, 437
n factorial, A14
M Nobel Prize in economics, 222, G1
Markov, Andrei, 436 Nonbasic and basic variables in linear programming, 273–4,
Markov chains, 436–77 279–80
Index I-7
Nonnegative constraints, in linear programming, 254, 309 Powers of transition matrices in Markov chains, 441–2
Nonstrictly determined games, G5–G6, G8–G9, G13–G14 Predictions, extrapolation for, 559
Nonstrictly determined matrix games, G19 Preliminary simplex tableau, 310
Notation Present value
summation, A3–A5 of annuities, 134–7
Not factorable, polynomial as, 582 compounding schedule and, 117–18
Not factorable in the integers, principal as, 105–6
polynomial as, 606 Prime numbers, 579
Notation Principal diagonal of a matrix, 167
double subscript, for matrix position, 167 Principal (present value) invested, 80, 105
for functions, 25–8 Probability, 372–435. See also Binomial distribution; Markov
interval, 534–5 chains
scientific, 592–5 Bayes’ formula, 413–18
for sets, 339–40 complement of events, 390–2
nth root radical, 596 conditional, 399–402
nth roots of real numbers, 595–6 decision making and expected
nth-term formulas, A8 value, 426–7
Number line, real, 568 equally likely assumptions in, 380–3
Numerical coefficients, 574–5 of events, 377–80
expected value of random variables, 422–6
O independent events, 405–9
Objective functions, 254 odds in, 392–4
Odds, in probability, 392–4 probability trees, 403–5
One-to-one functions, 85 product rule in intersection of
Open intervals, 534 events, 402–3
Optimal solution, in linear random variables and distribution
programming, 254 of, 421–2
Optimal strategies, in game theory, sample spaces, events, and, 373–6
G12, G19 summary of, 409
Optimal value of linear objective union and intersection of events, 387–90
functions, 254 Problem constraints, in linear programming, 254
Order of operations, 577 Product rule in intersection of events, 402–3
Ordered pairs of numbers, 541 Propositions, in logic, 330–3
Ordinate of points, 541 Provident Bank (OH), 101
P Pure and mixed strategies for games, G9
Parabolas, quadratic equations as, 51 Q

Pascal, Blaise, 372 Quadrants, of coordinate system, 541
Payoff matrix, G2, G4 Quadratic equations, 601–9
Payoff tables, 424–5, 427 Quadratic formula, 603–5
Permutations, 356–8 Quadratic functions
Piecewise-defined functions, 41–2 equations, inequalities and, 46–50
Piecewise linear functions, 486 graphs of, 50–3
Pie graphs, 481 Quantitative data, 482
Pivot operation in simplex method of linear programming, 280–4, Quartic equations, 607–8
311
Point–by–point plotting, 20 R
Point-slope form of linear equations, 547–8 Radicands, 596
Poiseuille, Jean Léonard Marie, 61 Raising fractions to higher terms, 585
Polynomial functions, 62–4 Random experiments, 373
Polynomials Random variables
factoring, 579–84 expected value of, 422–6
operations on, 573–9 probability distribution of, 421–2
Population mean, 490 Range, as measure of dispersion, 500
Population standard deviation, 502–3 Ranges of functions, 22, 25, 74
Position of elements in matrices, 167 Rate of change, 555–6
Positive real numbers, 568 Rational exponents and radicals, 595–601
I-8 Index
Rational expressions, operations on, 585–90 Solutions

Rational functions, 65–9 for e-system of linear equations, 267
Real nth roots of real numbers, 596 to equations, 531
Real numbers, 567–73, 595–6 feasible, 267, 278
Recessive rows and columns, in game theory, G14–G16 of games, G12–G14
Reciprocals of real numbers, 569 for i-system of linear inequalities, 267
Recursively-defined state matrices, in Markov chains, 441 of minimization problems, 298–300
Reduced matrices, 176–7 solution region for systems, 244
Reducing fractions to lowest terms, 585–6 Square function, 35–6
Reflections, stretches, and shrinks, 38–40 Square matrices, 167, 203–9
Regression analysis Square root, quadratic equations
exponential, 78 solved by, 601–2
linear, 557–1 Square root function, 35–6
logarithmic, 92 Standard deviation
Regression polynomials, 64–5 of grouped data, 503–4
Regular Markov chains, 449–51 significance of, 504
Relative frequency, 380, 483 of ungrouped data, 500–3
Relative growth rate, 77 Standard form, 266, 531
Richter scale, logarithmic, 84 State matrices, in Markov chains, 439–41
Right half-planes, 236 Stationary matrices in Markov chains, 448–51, 453
Root of a function, 63 Statistics, for data description, 483
Roth IRAs (individual retirement accounts), 131 Stochastic processes, 437
Row equivalent augmented matrices, 168 Stretches, shrinks, and reflections, 38–40
Row matrices, 167 Strictly determined matrix games, G2–G6
Russell Index of stocks, 119 Substitution method for solving systems of linear equations,
156–157
S Subtraction
Saddle values, in game theory, of fractions, 587–8
G3–G5, G14 of polynomials, 576
Sample mean, 490 of real numbers, 570
Sample spaces, events and, 373–6 Summation notation, A3–A5
Sample standard deviation, 503 Surplus variables, 309, 312, 321
Sample variance, 501 Systems of linear equations. See Linear equations,
Scatter plots, 558 systems of
Scientific notation, 592–5
Second-degree polynomials, factoring, 581–2 T
Security level, in game theory, G3, G12 Table method, simplex method and, 267–72
Selten, Reinhard, G1 Tautology, in logic, 334
Sequences Technology matrix, in input-output analysis, 224
arithmetic and geometric, A7–A12 Terms of a sequence, A2
overview, A1–A3 Transformation of graphs of functions, 36–8, 40
Series and summation notation, A3–A5 Transition diagram, 437
Set of real numbers, 567–8 Transition matrices of Markov chains, 439–42, 466
Sets Transition probability matrix, 437
properties and notation, 339–40 Translating graphs, 37–8
Venn diagrams and set operations, 341–3 Tree diagrams
Shrinks, stretches, and reflections, 38–40 of factorization, 579
Simple events, 373–4, 377 Tree diagrams, 403–4
Simple fractions, 588–9 Truth tables, 333–4
Simple interest, 105–9 22 matrix games, G12–G14
Simple outcomes, 373–4 Two-finger Morra game, G8,
Sinking funds, 129–31 G8–G9, G11
Slack variables for problem constraint inequalities, 267, 312, 321 Two-person zero-sum matrix games, G2
Slope
of a line, 544–6
U
as rate of change, 555–7 Unbounded solution regions, 246
Slope–intercept form of linear equations, 546–7 Ungrouped data
Index I-9
mean of, 491 Vertex form of quadratic functions, 50

median of, 494 Vertex of parabolas of quadratic
range of, 500 equations, 51
standard deviation of, 500–3 Vertical asymptotes of rational functions,
variance of, 501 65–6, 68–9
Union Vertical (y) axis, of coordinate
of events, 387–90 system, 541
of sets, 341–2 Vertical and horizontal shifts, 36–8
Unique solutions of systems of linear equations, 155 Vertical lines, 542, 544
Upper half-planes, 236 Vertical shrinks of graphs, 39
U.S. Army, 562 Vertical stretches of graphs, 39
U.S. Census Bureau, 92 Vertical-line test for functions, 24
U.S. Treasury Department, 107 Vertically translating graphs, 37–8
V W
Value of a game, G12 Weighted average, calculating, 492
Variable costs, 28 Written in factored form, positive
Variables integer as, 579
artificial, 309, 312
basic and nonbasic, in linear programming, 273–4, 279–80 X
decision, 254 x (horizontal) axis, 39, 541
equations in two, 20–1 x intercept of lines, 542
expected value of random, 422–6
independent and dependent, 23 Y
linear equations in two, 541–4 y (vertical) axis, of coordinate
linear inequalities in two, 236–40 system, 541
in polynomial terms, 573 y intercept of lines, 542
probability distribution of, 421–2 Yield, annual percentage, 120–2
restrictions on, 585
slack, 267, 312, 321 Z
surplus, 309, 312, 321 Zero factorial, A14
systems of linear equations in two, 153 Zero of functions, 47–8, 63
Variance, 500–1. See also Zero matrix, 466
Standard deviation Zero property, 196, 570
Venn diagrams, 341–3, 415 Zero-sum matrix games, G2
INDEX OF APPLICATIONS
NOTE: Page numbers preceded by G refer to online Chapter 10: Games and Decisions. Page numbers preceded by A refer to online
Appendix C: Special Topics.
Business and Economics Cost function, 28

Credit card accounts, 109, 141–142, 144, 149
Advertising, 83, 293, 323, 446, G33
Cryptography, 210–211, 213, 234
Agriculture, 288–290, G33
Customized vehicles, 243
as economy sector, 230
oil and, 229–230 Decision analysis, 426–427, 429, 435
tourism, 229 Delivery charges, 165
Airplane leases, 186 Dental insurance, 447
Amortization, 144–146, 150–151 Depreciation, 79
Animal food, 262, 264 Diamond prices, 557–558
Annual percentage yield (APY), 121, 124, 149–150 Doubling times for investments, 91–92, 94
Annuities, 128, 133, 135, 143–145, 149–150
Auto financing, 141, 144, 149 Earnings per share, 505
Automobile production, 60 Economy stimulation, A12–A13
Average cost, 71–72 Electricity rates, 44–45, 101
Electricity, natural gas, oil, 230
Bank promotion, G18, G25 Electronics, 165, 264
Bicycle manufacturing, 293–294 Employee evaluation, 452–453
Blending–food processing, 328 Employee layoffs, 366
Boat production, 186 Employee rating, 420
Bonus incentives, 412 Employee screening, 420
Breads, 307–308 Employee training, 69, 472, 476–477
Break-even analysis, 55–57, 60, 101, 164–165, 233, Employee turnover, 514
537–538, 540 Employment: resource allocation, 308
Business Energy consumption, 562–563
depreciation, 553 Energy production, 562
markup policy, 553 Equilibrium point, 95, 102
Business closings, 366 Equilibrium price, 161
Business leases, 233 Equilibrium quantity, 161
Equipment rental, 540
Capital expansion, 261–262 European Food Safety Authority, 412
Car prices, 539 Expected values, 424–425
Car rental, 446
Checkout times, 506 Federal income, 487
Coal, steel, 229 Fertilizer, 242–243, 262
Coffee blends, 165 Finance, 82
Committee selection, 386 Four sector economy, 230
Common stocks, 488 Furniture, 249, 260–261
Communications, 354 Gasoline prices, 487–488, 573
Commute times, 488 Gasoline tax, 498
Compound growth, 82 Gold production, 487
Compound interest, 114, 124–126, A13 Gross domestic product, 487
Computer control systems, 409 Gross receipts, 579
Concert tickets, 221 Guarantees, 514, 525
Construction, 102
Consumer Price Index, 538 Headphone sales, G30
Continuous compound interest, 82, 95, 116, 125 Health plans, 447
Corporate farming, G19, G25 Home construction, 293
Corporation revenues, 487 Home mortgage loans, 139–141, 145, 150–151
Cost analysis, 200, 552–553 Home ownership rates, 92–93
Cost equations, 549 Home values, 539
I-10
Index of Applications I-11
Homeowner’s insurance, 446–447 Physics, 554

Hooke’s law, 554 Plant food, 249, 262, 308, 323
Plant safety, 529
Incentive plans, 222 Playing cards, 362–364
Income tax, 187 Political advertising, 243
Individual retirement account (IRA), 125, 133, 150–151 Postal service, 487
Input-output analysis, 234 Preference survey, 528
Insurance, 344, 398, 429, 435, 451 –452, 457 Price war, G7
Insurance and expected values, 425–426 Price-demand, 28, 32–33, 44
Interest rates, 133–134 Price-demand and revenue, 29–31
Internet, 476 Price-earnings ratios, 498
Inventory value, 200 Price-supply, 44
Investment, 261, 578, G7, G18, G25 Product defects, 405, 417–418, 420
Investment analysis, 217–219, 234, G27–29 Product switching, 476
Investments, 94–95, 293, 324, 328 Product testing, 399
IRA (individual retirement account), 539 Production scheduling, 186, 221,
250–254, 261
Labor costs, 197–198, 200, 233–234, 242
Profit, 28, 33
Labor force, 456
Profit-loss analysis, 28–29, 60–61, 101
Labor relations, 412, 525
Public debt, 594–595
Lightbulb lifetime, 498, 518–519
Purchase price, 537
Loan distributions, 323–324
Purchasing, 183–184
Loan repayment, A11–A13
Loans, 105–106, 111–112, 137–139, 141–146, 149–151, 472 Quality control, 366, 391, 398–399, 412, 435, 506, 514,
Management, 353–354 522–524, 525
Management training, 514–515
Real estate development, 462–464
Manufacturing, 323, 328
Refinery–petroleum blending, 319–321
resource allocation, 293
Refining, 307–308
Market analysis, 434
Refund anticipation loans (RALs),
Market research, 348, 354, 370–371, 394–396, 398, 434,
112, 149
520–522, 529
Replacement time, 72
Market share, 457
Resource allocation, 233
Marketing, 472, 476
Retail and sale prices, 539
Marketing claims, 525
Retail and wholesale prices, 539
Markup, 200
Retail sales, 528
Material costs, 233
Revenue, 28, 33, 60
Mathematical modeling, 182–183
Mattresses, 243 Sail manufacture, 264
Maximum revenue, 54–55 Sales, 240–241, 525
Minimum average cost, 72 Sales commissions, 191–192, 540
Minimum wage, 354 Sales tax, 573
Money growth, 82–83, 101 Sausage, 165
Music, 262 Scheduling, 446
Seed costs, 242
Natural gas rates, 42
Service contracts, 447
Net sales, 563–564
Shipping schedules, 328
Oil refining, 324 Simple interest, 105, 111–112
Operating income, 564 Sinking funds, 133, 144–145
Outboard motors, 57–58, 61 Sound intensity: decibels, 95
Starting salaries, 488
Parking meter coins, 539 State sales tax, 571
Parking receipts, 221 Stock analysis, 506
Payday loans, 112 Store locations, G7–G8
Pensions, 472 Student loan debt, 144, 150, 498
Personnel screening, 529 Supply and demand, 95, 161–162, 164, 549–551, 554, 566,
Personnel selection, 366–367, 385–386, 412 607–609
I-12 Index of Applications
Tank car leases, 186 Human weight, 45

Taxable income, 187
Ticket prices, 234 Ideal weight, 561–562
Ticket sales, 539
Marine biology, 83, 102
Tire mileage, 59–60
Medical diagnosis, 515
Tour agency, G30
Medical research, 367, 371, 386, 420
Tourism, 498
Medication, 257–258
Trail mix production, 324
Medicine, 61, 102, 354, 367, 386, 399, 472,
Transportation, 261, 301–304, 324, 354, 370, 456
506, 526
Transportation, manufacturing, 229
Medicine–cardiograms, 435
Travel agent training, 447
Mixing animal feed, 324
Treasury bill (T-bill), 107, 111, 150
Mouse weights, 488, 499
Viewer ratings, G18, G25 Muscle contractions, 33
Water skis, 249, 260 Nutrition, 187, 246–247, 249, 262, 489, 506, 579
Welder training, 447
Olympic Games, 565–566
Life Sciences Patient recovery, 512–513
Agriculture, 95, 102–103, 526 Physics, 165
AIDS epidemic, 488–489 Physiology, 73
Air pollution, 595 Plant food, 187
Animal diet, 165 Pollution control, 261
Animal feed, 197 Population growth, 103, 488
Animal nutrition, 294 Pregnancy testing, 420
Atmospheric concentration of carbon dioxide, 559
Rate of descent–parachutes, 562
Blood cholesterol levels, 499
Body surface area (BSA) estimation, 556 Sound waves, 165
Boiling point, 553 Speed of sound, 562
Cancer screening, 420 Temperature conversion, 540

Cereal blends, 200–201 Tuberculosis screening, 416–417, 420
Climate, 565
Underwater pressure, 562
Diet, 72–73, 159–161, 222
Wildlife management, 540
Drug side effects, 515
Earthquakes, 165 Social Sciences

Ecology, 609 Archaeology: carbon-14 dating, 95
Epidemics, 515
Estimating fish weight, 64–65 Birthday problem, 391–392
Exponential decay of radioactive 14C, 78
Exponential growth of bacteria, 77–78 Cigarette smoking, 554, 563
Committee selection, 346, 367
Fertilizer, 165 Computer-assisted testing, 351
Flight conditions, 553 Consumer survey, 402–403
Flight navigation, 553 Consumer testing, 385
Forestry, 560–561, 564 Crime, 595
Freezing temperature, 564 Crime statistics, 102
Gene mutations, 457, 515 Demographics, 553–554

Genetics, 413, 429, 457, 457, 515, 526 Divorce, 73
Greenhouse gases, 489
Education, 222, 489
Health care, 529 Elections, 371
Heredity, 201 Enrollments, 345–346
Herpetology, 45 Entrance examination scores, 499
Human nutrition, 308, 323–324 Exit polling, 343
Index of Applications I-13
Family planning, 354 Police science, 420

Politics, 33, 201, 354, 367, 386, 399, 420, 458, 526
Grade-point average, 489–490, 499, 506 Population, 95
Grading on a curve, 526 Population density, 595
Graduate enrollment, 565 Population growth, 83–84, 187
Population shifts, 458
Home ownership, 456–457
Presidents, 499
Housing trends, 447–448
Psychology, 540, 165–166, 249, 262, 413, 472–473, 526
Immigration, 499
Rapid transit, 457–458
Insurance, 344
Reading scores, 506
Internet users, 83
Safety research, 45, 401–402, 609
Learning, 45 Smoking, 477
Learning curve, 83 Sociology, 188, 515
Learning theory, 73 State prisoners, 489
Licensed drivers, 563 Student retention, 442–443, 467–468
Life expectancy, 187–188 Study abroad, 489
Lottery, 430
Test averages, 201
Marriage, 73 Testing, 515, 525
Median age, 489 Traffic flow, 188, 234
Medicare, 103
Mobiles, 261 Undergraduate enrollment, 565
Mobile data traffic, 83
Voter preference, 435
Opinion polls, 515 Voting coalitions, 346
Opinion survey, 294 Voting patterns, 413
A Library of Elementary Functions
Basic Functions
f (x) g(x) h(x)
5 5 5
x x x
25 5 25 5 25 5
25 25 25
Identity function Absolute value function Square function

f (x) 5 x g(x) 5 u xu h(x) 5 x 2
m(x) n(x) p(x)
5 5 5
x x x
25 5 25 5 25 5
25 25 25
Cube function Square root function Cube root function

3
m(x) 5 x 3 n(x) 5 x p(x) 5 x
Linear and Constant Functions

f (x) f (x) f (x)
b b b
x x x
m.0 m,0 m50
Rising Falling Horizontal
Linear function Linear function Constant function

f (x) 5 mx 1 b f (x) 5 mx 1 b f (x) 5 b
Quadratic Functions
f(x) f (x)
a,0
Opens downward
k
k
x
h x
a.0 h
Opens upward
f(x) 5 ax 2 1 bx 1 c 5 a(x 2 h) 2 1 k
Exponential and Logarithmic Functions
f (x) f (x) f (x)
(0, 1) (0, 1)
x x x
(0, 1)
b.1 0,b,1 b.1
Exponential function Exponential function Logarithmic function

f (x) 5 b x f (x) 5 b x f (x) 5 log b x
Representative Polynomial Functions (degree + 2)

f (x) f (x) f (x)
5 40 40
x x x
25 5 25 5 25 5
25 240 240
x23 Fourth-degree polynomial Fifth-degree polynomial

f (x) 5
x22 f (x) 5 x 4 2 3x 3 2 9x 2 1 23x 1 8 f (x) 5 2x 5 2 x 4 1 14x 3 1 6x 2 2 45x 2 3
Representative Rational Functions

f (x) f (x) f (x)
5 5 5
x x x
25 5 25 5 25 5
25 25 25
x23 8 1
f (x) 5 f (x) 5 2 f (x) 5 x 1
x22 x 24 x
Graph Transformations
f (x) f (x) f (x)
g f h
5 5 5
g f
g
x x x
25 5 25 5 25 5
h
h
25 25 25
Vertical shift Horizontal shift Stretch, shrink and reflection

g(x) 5 f (x) 1 2 g(x) 5 f (x 1 3) g(x) 5 2 f (x)
h(x) 5 f (x) 2 3 h(x) 5 f (x 2 2) h(x) 5 20.5 f (x)
Finite Math Reference
Mathematics of Finance
Simple Interest
I = Prt A = P (1 + r t)
(interest I on principal P at annual rate r for t years; amount A = P + I)
Compound Interest
r mt
A = P(1 + i)n = Pa1 + b
m
(amount A on principal P at annual rate r, compounded m times per year for t years; interest rate per
r
period is i = ; number of periods is n = mt)
m
Continuous Compound Interest
A = Pert (amount A on principal P at annual rate r, compounded continuously for t years)
Annual Percentage Yield

r m
APY = a1 + b - 1 (APY, or effective rate, of annual rate r compounded m times per year)
m
APY = er - 1 (APY, or effective rate, of annual rate r compounded continuously)
Future Value of an Ordinary Annuity

(1 + i)n - 1
FV = PMT
i
(future value FV of n equal periodic payments PMT, made at the end of each period, at rate i per period)
Present Value of an Ordinary Annuity

1 - (1 + i)-n
PV = PMT
i
(present value PV of n equal periodic payments PMT, made at the end of each period, at rate i per period)
Logic
Negation Disjunction Conjunction Conditional
(not p) (p or q) (p and q) (if p then q)
p q ¬p p ¡ q p ¿ q pSq
T T F T T T
T F F T F F
F T T T F T
F F T F F T
q S p is the converse of p S q ¬ q S ¬ p is the contrapositive of p S q

Sets
Union Intersection Complement
A ∪ B = 5x x H A or x H B6 A ¨ B = 5x x H A and x H B6 A′ = {x x ∉ A}
A B A B A
(A < B is shaded) (A > B is shaded) (A9 is shaded)
Counting
Sets
n(A∪B) = n(A) + n(B) - n(A ¨ B) (number of elements in the union of sets A and B)
Permutations
n!
nPr = (0 … r … n)
(n - r)!
(number of permutations of n distinct objects taken r at a time without repetition)
Combinations
n!
nCr = (0 … r … n)
r!(n - r)!
(number of combinations of n distinct objects taken r at a time without repetition)
Probability
Union of Events; Complement of an Event
P(A∪B) = P(A) + P(B) - P(A ¨ B) P(E′) = 1 - P(E)
P(A¨ B)
P(A 0 B) = P(B) ≠ 0
P(B)
Bayes’ Formula
P(E 0 U1)P(U1) P(E |U1)
E P(E |U1)P(U1)
P(U1 0 E) =
P(E 0 U1)P(U1) + P(E 0 U2)P(U2) + c + P(E 0 Un)P(Un) U1
E9
P(U1)
P(E |U2 )
(U1, U2, c, Un are n mutually exclusive events whose union U2
E P(E |U2 )P(U2 )
is the sample space; E is an arbitrary event such that P(E) ≠ 0; Start P(U2) E9
similar results hold for U2, U3, c, Un)
P(Un ) P(E |Un )
E P(E |Un)P(Un)
Un
E9
Expected Value
E(X) = x1 p1 + x2 p2 + c + xn pn
GLOBal This is a special edition of an established title widely used by colleges and
universities throughout the world. Pearson published this exclusive edition
edition for the benefit of students outside the United States and Canada. If you
purchased this book within the United States or Canada, you should be aware
that it has been imported without the approval of the Publisher or Author.
Features
• Matched Problems accompanying each of the more than 300 worked examples to
help students grasp concepts and assess their understanding
• Explore and Discuss problems that help to introduce new concepts or build upon
a current topic
• Exercise sets differentiated by level of difficulty that allow instructors to craft
assignments according to their students’ levels
• More exercises in MyLab Mathematics, with approximately 3,340 assignable
exercises and new exercise types, including
• Additional Conceptual Questions to assess concepts and vocabulary
• Setup & Solve exercises that require students to show how they set up a
problem along with the solution
• Interactive Figures that illustrate key concepts and allow manipulation

(M1) (Raymond Barnett) Finite Mathematics For Business

Uploaded by

Copyright:

Available Formats

(M1) (Raymond Barnett) Finite Mathematics For Business

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(M1) (Raymond Barnett) Finite Mathematics For Business

Uploaded by

Copyright:

Available Formats

MyLab Math for Finite Mathematics

for Business, Economics, Life

Questions that Deepen Understanding

RAYMOND A. BARNETT Merritt College

MICHAEL R. ZIEGLER Marquette University

KARL E. BYLEEN Marquette University

CHRISTOPHER J. STOCKER Marquette University

Cover Image: Mark Nazh/Shutterstock

Pearson Education Limited

Chapter 1 Functions and Graphs . . . . . . . . . . . . . . . . . . . 19

Chapter 2 Mathematics of Finance . . . . . . . . . . . . . . . . . . 104

Chapter 3 Systems of Linear Equations; Matrices . . . . . . . . . . . . 152

Chapter 4 Linear Inequalities and Linear Programming . . . . . . . . . 235

Chapter 5 Linear Programming: The Simplex Method . . . . . . . . . . 265

5.3 The Dual Problem: Minimization with Problem

Chapter 6 Logic, Sets, and Counting . . . . . . . . . . . . . . . . . 329

Chapter 7 Probability . . . . . . . . . . . . . . . . . . . . . . . . . 372

Chapter 8 Markov Chains . . . . . . . . . . . . . . . . . . . . . . 436

Chapter 9 Data Description and Probability Distributions . . . . . . . . 478

Chapter 10 Games and Decisions (online) . . . . . . . . . . . . . . . G1

Chapter 10 Summary and Review . . . . . . . . . . . . . . . . . G31

Appendix A Linear Equations and Graphs . . . . . . . . . . . . . . . . 530

Appendix B Basic Algebra Review . . . . . . . . . . . . . . . . . . . 567

Appendix C Special Topics (online) . . . . . . . . . . . . . . . . . . . A1

Appendix D Area under the Standard Normal Curve . . . . . . . . . . 610

New to This Edition

New to MyLab Math

students with an opportunity to express their understanding of the topic in

Additional Pedagogical Features

■■ Diagnostic Prerequisite Test, located on page 17, provides students with a

David Busekist, Southeastern Louisiana University

Acknowledgments for the Global Edition

MyLab Math Online Course

NEW! Study Skills Modules

Developing Deeper Understanding

NEW! Additional Conceptual Questions

NEW! Setup & Solve Exercises

NEW! Interactive Figures

NEW! Note-Taking Guide (downloadable)

Graphing Calculator and Excel Spreadsheet Manuals (downloadable)

Student’s Solutions Manual (downloadable)

PowerPoint® Lecture Slides (downloadable)

Instructor’s Solutions Manual (downloadable)

Diagnostic Prerequisite Test

6. What is the leading coefficient of each polynomial? (A) 1 - 72 - 1 - 52 = 1 - 72 + 3 - 1 - 524

EXAMPLE 1 Point-by-Point Plotting Sketch the graph of each equation.

(B) Again we make a table of solutions—here it may be easier to choose inte-

To graph the equation y = - x3 + 3x, we use point-by-point plotting to obtain the

To each person, there corresponds an annual income.

Table 1 Table 2 Table 3

Number Cube Number Square Number Square root

Explore and Discuss 2

Functions Specified by Equations

DEFINITION Functions Specified by Equations

(B) Solving for the dependent variable y, we have

Matched Problem 2 Determine which of the following equations specify

THEOREM 1 Vertical-Line Test for a Function

The function graphed in Figure 7A is an example of a linear function. The

EXAMPLE 5 Finding Domains Find the domains of functions f, g, and h: